After selling Pebble in 2016, Eric became a partner at Y Combinator. For four years, he lived and breathed the unparalleled startup insight that only YC can provide, advising dozens if not hundreds of start-ups, and learning what makes some start-ups succeed and others fail.

Today, Eric’s back in the founder game with Beeper, a flexible, universal communication network, connecting you to 10+ different chat apps. While it may look like he’s out of the hardware game, it seems he can’t keep his hands off of PCBs and electronics.

If you’d prefer to listen rather than watch, we’re now publishing Atoms and Bits interviews as a podcast. Here’s this episode:

In this interview, Eric shared his perspectives on what he has learned from his experience founding Pebble and then advising start-ups at Y Combinator. While the best insights will come from watching the video or listening to the podcast audio, here’s a quick summary of the major takeaways:

• Speed is the key to survival. Most [hardware] companies operate too slowly, and it’s a universal truth that the market moves, and fast. You’ll need to iterate, fail, learn, move on, repeat, over and over again. In Eric’s words: "build it this week and ship it next". Migicovsky cautions on the excuses of waiting (on inventory, on outside decisions, on perfection), and urges founders to take action themselves to keep momentum. If what you’re building is truly transformative, it will outweigh the inevitable obstacles and errors.

• You can differentiate founders from the rest of the world based on the love for what they do. In Eric's words, it's the "ridiculous fact that they won't give up" that defines founders. If you listen to the people saying your project is bound to fail, you’ll never get started.The key is building something you need, want and can start using right away- as you are the first to test it you are able to fall in love with it and communicate that passion to your first group of believers. Do not forget to never stop defining a vision for the future of your product and company.

• You’ll always go farther with a team. Seek outside expertise, forget your assumptions of business as usual. Embrace Silicon Valley as a mindset, not a geography, and you’ll find those who share your enthusiasm, urgency, and belief that everything is possible. Do not miss Migicovsky’s advice for a successful application to Y Combinator. Among other things, he stresses the point of succinctly articulating what your product is, why it’s important, and why you’re the right person to bring it forward. Oh — and make sure you can clearly illustrate the impact you’ll have on the world.

Share

Avidan is the Founding Partner of Root Ventures. Before founding the deep tech VC firm in 2013, he designed industrial robotics for the Food Network's kitchens. Before that, Avidan was the CTO of CIM Group, a $15 billion investment firm, where he focused on industrial internet investing.

Root Ventures is an early stage seed fund that invests in highly technical businesses led by technical founders. Root recently announced their third fund, bringing another $150M to bear to drive the tech industry forward.

Audio Only 👇

This interview is about an hour long, and is best consumed in video or audio. That said, if you’re looking for the Cliffs Notes, read on, and further down we’ve included a transcript of the interview.

Atoms & Bits will alternate between essays and video interviews like this one; tell us what you like and dislike as we’d love the feedback for future issues and episodes.

Key Takeaways:

• Investing in “deep tech” is about investing in technical risk. Every early stage business carries significant risk; that risk is what creates the opportunity for significant return to the investors. Businesses that take on extremely technical challenges face a set of technical risks (essentially that they might not be able to pull off the tech to make their product work) that represent opportunity for the right kind of investor that knows how to evaluate that risk and the founders’ ability to overcome those risks.

• Hardware/software businesses represent a subset of those “deep tech” businesses that carry unique challenges. Bringing a physical product to market as a start-up is challenging, and it requires the right founders who will survive the early technical challenges and the right investors who know how to help the founders navigate the challenges they will face.

• Hardware/software businesses often require unique business models that are unfamiliar to most venture capital investors. Many hardware/software businesses are not “venture backable” — they don’t have the kind of financial flywheel that generates the exponential growth required to appeal to VCs. That said, some hardware/software businesses do create exponential growth — but often through business models that may be unfamiliar to a traditional VC. Those businesses then have to educate investors to help them understand how their business model will create the kind of returns that a venture investor is looking for.

Transcript:

Intro

Avidan (00:00): ...to hone your pitch and think through the business model. The whole way Silicon valley works is a pay it forward attitude of everybody willing to get on the phone and share whatever they think. By the way, if you just want to get on the phone with me to get advice, that's easy. If you're not asking for money, it's a much easier... And that's how everyone in Silicon valley is like. Honestly, if you saw me at a cocktail party or you saw me at a coffee shop and you wanted to come up to me and chat for a couple minutes and hear my thoughts, I'm not shy. I will tell you what I'm thinking and that's great.

-

Zach (00:33): Hi, my name is Zach Supalla, and this is Atoms and Bits. We're here interviewing Avidan Ross of Root Ventures.

Avidan (00:39): Absolutely. Thanks for having me. So I'm the founder of Root Ventures. We're a[n] early stage seed fund, based here in San Francisco and we invest in engineering enablement. So what we mean by that is we think technical risk is worth taking. There are a lot of people out there who believe venture capital has always funded people taking the biggest technological risks. And the truth of the matter is over the last 10 to 15 years, there have been other opportunities to invest that aren't really about engineering risk. It's about go to market risk or it's about internationalization risk, or it's about a brand and building brand. And at the end of the day, we believe that funding deep technical risk is a risk we want to take.

So the way that we manifest it is we invest in people who are building tools for engineers, and we invest in engineers who are building companies. So the two types of companies that get created are ones where there's a lot of technical risk in getting the company started. Things like hardware, robotics applied AI, synthetic biology. And then we invest in people who are building tools to make those companies easier to build. So we end up investing in companies like electrical engineering software, mechanical engineering software, data science software, a lot of things like IOT development platform. So our relationship was that we were the first investor in Particle and we have continued to invest in this space, but it all comes down to, we believe that engineers need the funding and I'm happy to share with you that, that's a chip on my shoulder, because my background is an engineer and we can dive into that if you want. I don't know which order you want to go in.

Zach (02:14): So just for the audience's purposes, Avidan invested in Particle like a thousand years ago. What, eight years ago?

Avidan (02:20): Eight, nine years before COVID. Nine year BC. No, I'm joking. Nine years ago, 10 years ago.

Zach (02:29): Something like that. So Avidan and I have been doing this for a very long time. So all the questions I'm going to ask him, I've already asked him and we've already talked about all the stuff a thousand times.

Avidan (02:36): You're a part of half of the story and I'm part of half story.

Zach (02:39): Yeah, exactly. So like this is all going to be just like all the stuff we've been talking about for many years. But so like let's go back before that. Yes. You found your way into VC. How did you find your way into VC? What caused you to get into this?

Avidan (02:54): Yeah, I've heard of a lot of people taking a pretty direct route. People come up here and they study business and they go to business school here and they work at Google or they work at Apple and they work at Andreessen Horowitz and work their way into starting to invest in tech companies. And for me it was totally not the path.

So like everybody at Root, I have an engineering background. I studied computer science and embedded systems. And during the first dot com boom, I was writing embedded firmware and network application software at Excited Home, which was a company that basically invented high speed broadband. So we were building cable modems. But long story, short dot coms crashed, I bought a round the world plane ticket. And when I got back to the US, the dot coms were still crashed. And I found my way into building tech for a private equity firm in Los Angeles. And it was a small firm doing a lot of real estate and infrastructure investing.

And when I say infrastructure, I don't mean like data center infrastructure. I mean water, wind, solar smart grid. And I was very lucky to be in the right place at the right time. We grew tremendously quickly. And as the CTO, I was overseeing how we deploy technology, both in companies we were purchasing and how technology transformed that business, as well as looking at technology that transformed the way that we operated as investors.

And my revelation was that so many of these huge industries we were looking at could be transformed. If someone was willing to take that technical risk. If someone was willing to invest in fundamental technology that applied to pick your industry, construction, agriculture, mining, waste management. And I had this working assumption that, I was in Los Angeles at the time, and I was like, "well, in Silicon Valley, they're clearly investing in that", but this was 2009, 2010 and all the rage was, technology has been pre-built for you. You have Apple building the iPhone and iOS store and platform. You have a lot AWS basically scaffolding everything for you-

Zach (04:58): So a lot of social mobile, local time.

Avidan (05:00): ...Yeah. Mobile, local, slow, mobile, local social time.

Zach (05:03): It was that time.

Avidan (05:05): And I started getting approached by engineers who were saying, "I want to build a company, but I need someone who's willing to take that technical risk". And so I went to my partners at the firm and they were very supportive, but they said, "look, we can't invest as a multi-billion dollar private equity firm. We're not looking to take technical risk. We should invest when they're already grown and they have scaling risk and internationalization risk. Why would we take that technical risk?" And when I heard them say that, I realized that was my unique opportunity that I could take the technical risk.

So I gave them a year of non-compete and I did all sorts of wacky stuff. I basically told them, "I'm not going to invest for a year, but that means I'm going to travel around the world. I'm going to write a book about the best coffee shops around the world. I'm going to build crazy robots for the food network," which was a fun non sequitur. Although I will tell you the show never made it to air because my cohost lost his eyebrows in the first episode, because we built a pizza oven that cooked pizza in 45 seconds using robotically, charged after burners. And we didn't build a safety switch on it.

Zach (06:18): Does that pizza oven still exist somewhere?

Avidan (06:19): It eventually got taken apart. I didn't have space to build it, so my parents offered their backyard. And my father, I remember the day we were about to film the final shot of this crazy oven. And he shows up and he looks, he said, "did you guys take out insurance on this?" And I looked at him and I said, this is the wrong time to ask that question, because we're rolling. We're in a film right now. I was like, "mom, can you get rid of dad? He's asking questions about the insurance." But it was a lot of fun and it sort of reminded me how much I love building things.

And the segue into venture capital was that I started meeting engineers who I kept on leaning into this. I want to help any technical founding team figure out how they go to market. Figure out how do they sell to customers? How do they figure out their go to market motion, but also what technology is worth de-risking, what's worth building, how do build out your business model? And for me, the chip on the shoulder was I was the technologist who was always told technology is about cutting costs. Not about driving growth and transforming your business.

And so I put my money where my mouth is and I started angel investing. But before I could write any check on my own friends of mine said, we want to invest with you. We know that technical risk is worth taking, but we're not ones who can assess that technical risk. And that transformed into a bunch of companies that I invested in minor syndicates. And now we're in our third fund, we've invested in a bunch of special purpose vehicles.

We've - we have about 350 million under management at this point. So we are very much focused on those early stages and it all came from a failed Food Network episode where my buddy lost his eyebrows. Which I'm in retrospect, thankful for the lack of safety switch on that. Because I could have just been a, no offense Guy Fieri, but I could have just been frosted tips and out there building crazy cooking robots.

Zach (08:23): I assume his eyebrows grew back.

Avidan (08:25): Oh he had plenty to lose. He was all right. I think his girlfriend at the time was like "you know? They're trim. I actually..." When they grew back, she was like, "that length, keep them that length."

Zach (08:40): So, okay. You go from the private equity world into the venture capital world after a segue out and then in, with a focus on deeply technical problems and engineer solving them. My experience, this is about when you and I meet.

Avidan (08:57): Yep.

Zach (08:58): And the thing that I will always remember is that you showed up, I gave our whole pitch to a room full of investors about what we were doing, and you showed up with holding in your hand, a pair of our competitor's products. And we hadn't built our thing yet, so of course you didn't have ours yet, that's Kickstarter campaign at the time. But you came and you said "I know what you're building. I know what the alternatives are. I know there's a problem here. This is really interesting. I want to talk."

Avidan (09:23): And the alternatives are, I think my words were, "I don't think-"

Zach (09:27): "These products suck."

Avidan (09:28): Yes. I think it was, "these products suck." I think I said, " I don't think you're the first person building this. I think you're the first person building it right." Right. And it was this idea that I had seen so many big industries that we were trying to transform with connectivity at my old private equity firm. Right. And as the CTO, I ran the R and D group to try and figure out what would work. Most of those companies I was holding are now bankrupt or sold for pennies. So I'd rather not list out the names of all the modules I was holding.

Zach (10:02): Those don't exist anymore.

Avidan (10:03): Yeah. But the idea was that those modules weren't thinking about how businesses could be transformed. They weren't treating the engineers who were developing on them as first class citizens. Instead they said, "we've got this solved. You're going to sign up for our white paper. You're going to get access to our development kit, but we don't trust you with the keys. We're not going to let you see inside. We're going to give you a command list and you may use this command list." And you were like, "no, no, your thing has a bug in it. I found the bug. Somebody listen to me, the thing is broken." And they're like, "no, no, no, we've got this." And it was just a very old guard method of building hardware. So I do remember, I think you've subsequently described it at a party or two as a fist full of microcontrollers in one hand and a fist full of cash in the other.

Zach (10:52): Right. Exactly.

Avidan (10:53): And it was more like a wire.

Zach (10:55): But yeah, this doesn't actually work with bags of money like the emoji-

Avidan (11:00): Only in crypto. Yeah.

Zach (11:02): So, okay. So your thesis, now when we met, aside from the fact that you understood the problem, which was unique. I was talking to a bunch of VCs who I had to explain the fundamentals of, well, why did anybody build an IT product? Why does somebody need a product like ours in order to be able to do that? You just knew. And I think that when it comes to investing in technical founders and technical products, just knowing and understanding those problems is a big part of what puts you in a position to make the right bets.

Zach (11:32): You also were at the time, one of the few investors who didn't run screaming from something that had a physical product associated with it. We're showing off circuit boards in a pitch deck and most investors are like, "Nope, that's not a thing that I'm interested in", you leaned in. How do you think about that? And especially that was what, 2012, 2013, something like that, that's 10 years ago. World's changed a lot and I'm sure your perspective on it has become much more refined. How do you think about why are you not scared of it? Let's frame it that way.

Avidan (12:12): I am a little bit scared of it, but it's scared in the right way. It's understanding that it is a high risk endeavor, but it's high reward. If you do it well, you build something that's far more valuable and far more difficult to recreate and far more difficult to switch out. And I kind of fall back to the roots of venture, which is you should take on the risk, right? Risks are worth taking, but you need to make sure that the upside is commensurate with the risk you take. And I think at the time IOT... I actually have a real hard time with IOT as a three letter acronym. Because I think that there was a time and place, and it wasn't long after we met, that IOT meant stuff a chip in it. Grab your toaster oven, make it an internet thing.

It's already a thing, make it now internet. And I think no one was really contemplating how industries could transform with connectivity. Because really at the end of the day, if you think about what does it mean to be a venture backed hardware business, or what does it mean to be a company that's enabled by connectivity? It is this idea that like you are doing something at the edge.

You have a piece of equipment, you have a robot, you have lighting, you have air conditioning, you have whatever it might be. And you benefit from not being alone. From not being an isolated asset from the benefits of infinite compute in the cloud. The access to worlds of information, access to information about what is the current cost of electricity at this exact moment? What is the forecast of weather in the next six hours in the next six days? What are the other devices on the network doing? And then most importantly, it's that you can impart change. So it's not just reporting out information, it's then getting that information back.

And I think that there was a moment in time when people said, "no, we need to just take every thing, and think, can we sell it more if it's connected?" And it destroyed the industry. I mean, the IoT industry spent these years where everyone was banging their head against the wall because everyone just assumed "oh, add compute and intelligence, and now it's a venture backable thing."

And we think about it totally differently. So we think about it as, does connectivity make this business sticky, scalable, generates a larger TAM, grows the overall market. Make something possible that was never otherwise possible. And for any listener out there, who's trying to understand what does it mean to be venture backable, what it means to be venture backable is venture capitalists are actually pretty terrible at investing. Most of our deals go to zero. And we don't know which of the deals is going to succeed.

But when it succeeds, it needs to make up for all those losses. And so what we look for is only companies that have the potential of becoming extremely big and extremely quick. And so you might look at a venture backed deal and be like, "that's a dumb idea." But the question you have to ask yourself is what if it works? What if it works? And if it works, can it become so big? And the reality is venture is not the core investment method for most successful businesses.

When we think about, what does it mean to be a successful business, you can walk down the aisles of target and Walmart and any retailer, or even on Amazon for that matter, there are successful businesses and there are successful products, and most of them weren't venture backed. At the end of the day, they're building product. And that product is great.

So we're not scared of companies that are building something that's technically complex, or has a high risk. But what we make sure of is that the upside and the exit of, what if it works, what if this hardware works? How big is the potential upside? Is the market even more massive than we could have ever imagined? And that's how we make those assessments. I'm happy to go into, if you want to talk about individual detailed ones, but I think that at the end of the day, a lot of investors are looking for risks they understand.

And hardware risk is one that there are very few VCs who can understand the risk of, what will it take from get to get from point A to point B. And I think I'm sitting here, but the reality is our secret weapon at Root Ventures is that there are five of us. And one of whom is Chrissy Meyer. And she was the EPM on the iPod, on the Apple Watch, she ran hardware at Square, she was the founder of actually an IOT-

... hardware at Square. She was the founder of actually an IOT accessory business called Pearl. So when the two of us work on things, she's able to really understand the risk of supply chain, chips, chip shortages, how the hardwares get manufactured. And I'm thinking about the customer base and how does the customer base's business change and all of those things. And unfortunately, it's a risk that most VC firms don't have the stomach for.

Zach (17:25): So that I think is... I mean, one of my theories of venture capital and hardware is that there is a framework for what a good business... Like a sort of hardware, software combo business. Right? I mean, I'll start with by saying I think most businesses that are just pure hardware products, that are like a physical widget that you sell, by default are probably actually not venture backable.

Avidan (17:46): I agree.

Zach (17:47): Right? There are going to be exceptions to that, but the fundamental economics of selling a product that way do not match well with VC. So therefore, we're really just talking about hardware, software businesses. Businesses that have, yes, a physical product, but also software and create value through that combined effort. And that the hardware software businesses can be venture backable or cannot be just like a pure software business can or cannot be. Not just because you make a software product doesn't mean it's venture backable either.

And so then it's a framework and you say, what is venture backable? What isn't? And I think most VCs have a framework for a SaaS product. They know what to look for. They know what questions to ask, what metrics to look at, to be able to say this is a good one, or this is a bad one. Most investors who don't do any hardware, software investing just don't have that framework for hardware, software. So everything looks the same. And you are one of the few who has done enough of these things to actually have an opinion on what good or bad looks like. So actually before we get into that, because I want to understand what your framework is, but before-

Avidan (18:55): Oh no, I was like the framework, it's the golden rule of seven, but no follow up question.

Zach (19:01): Yeah. It's like, where's the white board? We need a whiteboard. Okay. Before we get there, I'm curious actually in the last decade you have become one of the relatively finite number of early stage investors that like... If you're starting a business that has a physical product and you're like, "What VCs are not going to like run screaming from this?" There's probably like a half a dozen and you're one of them.

Avidan (19:27): Yep.

Zach (19:27): Which, my guess means that you get a lot of phone calls from people who make widgets.

Avidan (19:31): Yes.

Zach (19:31): Right? I mean, first off, like how do you feel about that? Are those good phone calls? Is that good pipeline or is that-

Avidan (19:41): I mean, from like a personal perspective, I love it because I still love product. My wife hates it because our house is filled with me purchasing all those widgets because I can't tell you how many times I say absolutely customer, not an investor. I love what you're building and I want to have it in my home or in my office or part of my life. Because to be honest, everyone who's building hardware, whether it's hardware plus software, the moment you're building hardware, you have to have a certain passion about what you're working on because it's so difficult. You have to care so much and it becomes so clear so quickly that you are climbing up a serious hill. It is not a weekend project. It's not a quick wire frame or an app that is just like a quick proof of concept that might be a lottery ticket.

There is no quick and dirty hacker camp of just like, well let's build the product that we're going to now go sell. You have to care a lot about it. And so the people who come and share with me what they've built, they come along with a passion. They come along with a care and a belief in it. And I try every time to make it abundantly clear that whether I invest or not is not an indication of whether or not what you've built is awesome. Because there are a lot of products I love in my life that are not built for a venture investment.

And so I think what it comes down to is I love hearing those pitches because I genuinely get excited for the people who are attached to those. And then I try my best to just help them, even if it doesn't mean investing, which I know is really hard. Come for money, get advice. But I try my best and I try to be really helpful. And what I will say is like every once in a while we get pitched a really interesting product and the person surprises me pretty massively when they say, "Oh, this is a Trojan horse for something huge." This is a hardware product. And on the surface you might think, oh, that's just a hardware product. And they say, " No, but it's a hardware enabled blank." And that's where things start to get interesting.

We are a hardware enabled services business. We are a hardware enabled software business. We are a hardware enabled platform business. We are a hardware enabled recurring revenue business. We are a hardware enabled data business. And when people start digging into that, that's when I really lean forward. And I say, okay, okay, let's talk about the what if this is successful. And the fun part about these businesses is that there are multiple stops of revenue along the way, but we always know what the biggest version of that business is. And that's when I really get excited, when people are able to describe how does this become a multibillion dollar business or even larger.

Zach (22:50): Okay. As you describe what you're looking for, it sounds like you're very much focused on the end state which is like, what will this become? There is a higher degree of idiosyncratic risk associated with hardware businesses than the equivalent software business just because it is harder. Right. I think we'd `all agree, like the whole hardware thing, like it is true. How do you think about that? Because your thesis is we're focused on technically challenging problems, is that just part of the game or is there an exercise that you have to go through to validate, well, yes. Okay. Big vision. It's going to be hard. Are you still going to say no to some of these deals because it's going to be too hard?

Avidan (23:30): Yeah. It's a great question. We've talked about this a bunch and I think we can't delude ourselves to believe that there are people who think like us at every stage. So the really hard thing is like you listed off that there are six early stage folks, and it's probably more like four who are willing to take on that really-

Zach (23:51): These days, it's probably four.

Avidan (23:52): Right.

Zach (23:56): It was once six.

Avidan (23:56): Yeah. And then the last-

Zach (23:56): It was once like 15 and then it was six and then it was four.

Avidan (23:58): Yeah. And it's totally fair that that number has dropped. But the bigger question is how many people in the series A and the series B want to take on a level of risk where it's still a triple black diamond deal. So what we look at is we say, okay, each round you get two. Eventually, you're like, well, there are very few... Let's go to all the way to the end, the public markets.

Zach (24:31): Yeah.

Avidan (24:31): Right? The hypothetical full end. Are they willing to take on a company that says we still have a lot of technical risk to de-risk. We have a lot of technology we still need a de-risk. And it there's like zero, nearly zero companies out there. Elon Musk is the only person who can sell technical risk on an ongoing basis. And otherwise you saw what happened to the specs. The specs tried to sell that vision. And it was a very short lived experience in the public markets. And what I believe is that you have to realize that your goal as a deep tech company is to eventually just become a tech company. You want to take on the risk that you know you can retire. What you want to do is say, okay, if this company can... We like to look at it as a J curve, right?

You drop under and you're making no money. You're building technology. You're doing the really hard stuff. And you might not ship a product for two years. But after two years, when you start shipping, you're going to be generating revenue. You're going to have a customer acquisition cost. You're going to have a sales team. You're going to do all these things. And sometimes it's two years, sometimes it's three years.

But our take on it is how much money will you need to raise before you can be a company that starts to pattern match in a way that a broader universe of investors can assess the opportunity? And so that's what we love to do as a firm. What we love to do is invest a couple million dollars in the early stages. And we have a couple friends that like to do a little bit after us or with us. And we know that a company can spend anywhere from two to seven or $8 million and then emerge as a company that starts to look like a venture backable business to a very broad audience.

Zach (26:23): Yeah. That's interesting. So as a seed investor, you have to think about you're not going to be the last money in.

Avidan (26:30): That's right.

Zach (26:30): Or if you were, that's probably not what the intent was. If you's trying to build businesses that are going to be multibillion dollar businesses, then the seed is going to be the first of five stages investment before you get there. And so that company has to be viable through each stage and then lead to some kind of exit, whatever. And so you have to evaluate a business for its viability for later stage venture funding. I mean, I imagine every early stage investor has to do that. Do you think you have to think about that more and differently because you're investing in deeply technical products than the equivalent early stage investor who's investing in traditional enterprise SAAS products?

Avidan (27:09): Yeah. Yeah. I think it's different because the risk truly goes away. There is a binary switch when you say the hardware functions. It passes, it turns green. The light turns green. The hardware has shipped. The robot functions. The robot is harvesting the strawberries, the robot is making the pizza. The machine in biotech went through a clinical trial. And in the clinical trial, the performance was amazing. That is a switch. That is a moment when you say the largest and most pronounced technical risk is complete. The rocket launched. The rocket landed. These are big milestones that in a lot of other investment worlds, it is many, many, many small steps.

Zach (28:01): Yeah.

Avidan (28:02): And the person who's not doing seed rounds of FinTech, they'd be like, well, it's actually all along the way. It's minorly decreasing risk throughout the entire thing. It's not a big moment. There's not a shipping moment. There's not a delivery moment. And I think that for a lot of those companies, that's why they love investing multi-stage because there are all these different entry points for them. There are all these different times and they just assess the risk at each juncture. But with deep tech there oftentimes is the like does this thing do what we hoped and dreamed it would be able to do? And that doesn't mean it's the only risk of the business.

This is why it's so difficult is even after you've de-risked that part, you still have these other risks you have to figure out. You got a branding risk. You got to go to market. Does your software work? Do you brick all your devices with a bad over the air upgrade and realize that we thought it worked, but they picked the wrong platform to develop on. They picked the wrong chips. They built it in house when they should have partnered. And those are the things that you start to figure out.

Zach (29:09): Yeah. We see this at Particle with our customers where in a lot of ways, our business operates like the Twilios, the Stripes of the world where you say like, okay, we're building a platform upon which our products are going to build products. Our success is tied to their success. If they expand their roll out of their product, we expand with it and their success is our success. And that's true of a business like ours. It's true of a business like Twilio or Stripe, or a bunch of companies that sell like dev tools into software products.

The difference that I see in our business is that our customers don't grow along a curve. They grow in stairsteps. And so somebody will be little and they might be little and then disappear, or they are little and then they go from being little to big and it almost happens overnight. And it doesn't happen in this very smooth way. It has that sort of binary, stair step pacing to it. Now, is that true? Because you do a bunch of deep tech stuff that's not hardware, right? Like CAD software, things like that. Is that dynamic also true for software deeply technical products or is that really just kind of a hardware thing?

Avidan (30:16): Oh, it's totally different for software. For software, you do start to see these people are picking it up. I mean, the graph can still look very hockey stick. But it's not like in hardware where you have a, well, we just got a huge contract or we are now able to step up to a new contract manufacturer where minimum order quantities are this high.

And we have this huge seasonal purchase.

With software, what you see is momentum. With software, so a lot of our companies in the electrical engineering world or some companies like anthropology and mechanical engineering. What happens there is a lot of users are using anthropology software to design their mechanical parts. And it becomes viral internally. So you see the part and you say that... I mean, so for, for context, anthropology does implicit modeling. So generative design, oftentimes for advanced manufacturing, like 3D printing or advanced CNC milling. So the parts look wacky. They clearly were not designed with like pen and paper and they were not designed using traditional CAD drafting tools.

Zach (31:22): Right. I'm reading the expanse right now. And I imagine the things that come out of anthropology look like the spaceships in the expanse.

Avidan (31:30): That's right.

Zach (31:30): When I'm reading the book.

Avidan (31:31): That's right.

Zach (31:32): I haven't seen the TV show. So I assume that they actually-

Avidan (31:34): And I don't know, I could see it in your eyes that you get it. I have no idea what you're-

Zach (31:40): It looks like alien stuff.

Avidan (31:41): Exactly, exactly. Because there's an algorithm that's generating the ideal design for light-weighting, for performance, for tensile strength, for whatever your parameters are. And as software users, as users in an organization are using it, their output is seen by others. They say, well, how do you design that? Then they get a training on it. And you get more and more licenses. And people are just adding licenses within an organization because they've seen the value of it. And so with a lot of software, you just see the adoption and you can see companies that are doing product led growth. But at the end of the day, most of those companies are one off until you get to enterprise sales. At which point, it starts to look a little bit like stair stepping, because there's a big enterprise contract that comes in.

But even then, it's not the same as hardware companies that live and die by Black Friday or seasonal spending or their factories that stand up a certain amount or the minimum order quantities that they have to hit. And then we also see companies that are building hardware that look nothing like that stair step. And oftentimes those are companies that are hardware enabled, but they're hardware enabling a verticalized business. So I'll give you an example. We have an investment in a company. It's called Oma Fertility and they run fertility clinics. But that's not the core technology. The core technology is their equipment is machine learning enabled to select the ideal sperm candidate in a fertility treatment. Now that is a piece of hardware that unlocks in a totally new type of business that is extreme.

The idea is that if you can improve outcomes by a very, very large percentage, and I won't talk about those percentages because that's part of their... What's amazing about that business is how big of an unlock it is. Now all of a sudden you're offering... You look to the world like a fertility clinic. But at the end of the day, you are hardware enabled. You have built a piece of technology that is connected, is intelligent. It's actually robotics at the end of the day. There's articulation that's occurring. And at what we made, we go back to that, like what is that big risk that we had taken on? The big risk was does machine learning improve the selection and outcomes in fertility treatments?

Election and outcomes in fertility treatments. Let me tell you, when we saw the numbers, we knew we had flipped a switch. That's unique in that type of business. That's what I think is unique in deep tech, but the growth there is going to come down to something that you would've never guessed is a problem, which is real estate, the construction of fertility clinics, what it takes to build these clinics out. Then, there's only a certain number of people that go in for fertility treatment at any given time. It's not like everybody gets the new iPhone at the same time and. There's a release of a phone and you see that ladder step. At the end of the day, there's a certain number of people who are looking for treatment in any given geography.

Zach (34:45): Right. It just turns into sort of a services business that grows in a different kind of way than that stair step.

Avidan (34:53): That's right.

Zach (34:53): Right.

Avidan (34:54): That's right.

Zach (34:54): Right. But you have to get over that initial milestone, delivering the product, proving that the product works, so there's a stairstep at the beginning.

Avidan (35:00): That's right.

Zach (35:01): You're coming in at the beginning, so you are always seeing a stairstep. That's the technical risk that you're taking on. You're then thinking about what does the next investor want to see. You're painting a picture where you're saying there's some milestone that has to be hit in order for this ... This is currently a seed stage investment, this is not a series A investment. I imagine if you're doing this in pure software world, you're thinking about it as, "Well. I want to see the metrics develop," but at the end of the day, if you have a 140% net revenue retention at series A, then in order to have a good series B, you just need to not lose that. You need to maintain 140% net revenue retention at series B, which if you keep that up ...

Avidan (35:44): Is that easy?

Zach (35:45): I'm not saying it's hard, but I'm just saying keeping up those numbers as hit higher stages is very hard.

Avidan (35:52): Yes.

Zach (35:53): You're fighting forces that are going to draw those numbers down, but you're trying to keep them as high as you can decline a little bit maybe, try and keep those figures up. But you're talking about it's a totally different staging, it's a totally different way you match the stages of VC.

Avidan (36:08): Yeah. Absolutely.

Zach (36:10): Then you're thinking, what milestone do they have to hit for the next person to be interested in investing in this business? Do the entrepreneurs that are pitching you, are they thinking about that way too? Or are you working with them to figure out let's think about how we would get you ready.

Avidan (36:25): Yeah. I think it's more that we work with them to help them understand, look, you're going to spend two years down in the fox hole building technology. You'll get no credit with revenue. You'll get no credit with metrics. It is okay for you to go down into deep R and D land to de-risk something majorly. Then, what we will do is help you message that to the next stage investor so that they understand that that big risk, that risk, was de-risked and that is the major risk. Because now what happens is, the business you have has an unbelievable competitive advantage. Not just a two year head start, but a two year head start where you were funded to work on really difficult technology, something that maybe you're the only team in the world that can do.

Now, if you are correct in your assumptions about the market, you might be an N of one company. You are category defining, category creating or category killing. The idea that you can come out with a technology enablement that allows you to charge half of what anyone else can do, and because of your automation, you're in a situation where you're saying, "I can charge half of what everyone else charges. I get the same margin." Therefore, all the people who are doing this without the hardware, without the robotics, without the AI, without whatever it is that you've built, what you spent the two years building, that whole business disappears and possibly your ability to charge less means that the total market size is huge.

You step into an investor who usually is taking pitches about X million ARR net revenue retention. Now, instead, they're looking at it saying, "You are the only company I've ever been pitched that can do this. What I need to believe is that that risk that has been de-risked leads to a business that moves faster and is safer or more protected or has a larger potential outcome than these other software companies, where the first two years, they figured out a way to get to 2 million ARR. Next year they're going to do 6 million in ARR. Then who knows what happens after that?" The difficulty of getting started is a feature, not a bug, in that situation. Right. The difficulty of getting started and someone having funded you to work on the most technical challenges means you become a different type of company when it's time for you to get revenue and scale.

Zach (38:49): You are one of the few investors who does this stuff with regularity. You have expertise in being able to separate good hardware, software businesses from bad ones. There's a small handful of investors who do that. The rest of the VC world still occasionally makes investments in things that have hardware with them, but they don't have the experience and knowledge necessarily to have a good framework of like what to invest in, what not to invest in. Where do you see the typical VC investor making bad bets? What are their false positives? Where do you see them not making bets they should be? Where are the false negatives for the typical VC against these kinds of businesses?

Avidan (39:25): I mean, it's a great question because literally the industry is piled with failures. If you think about it, the number of people who are currently investing in hardware, it's a much smaller number, and for good reason. Most of the VCs who dabbled in hardware and most of the hardware companies that even existed, don't exist anymore.

The hardware investors, the people who are okay with a little hardware are like, "I don't want to touch hardware again." How do you bundle up all the failures, all the mistakes made? A lot of it was this belief that you could sell a consumer on a piece of connected electronics, and you didn't have to think about the business model change. You're like, "Oh, they need smart lighting." You're like, "Great. And then what?" They're like, "Well, then it's the full ecosystem and platform for all their connected devices in their life. You will have smart lighting that leads to smart buildings that leads to smart cultures. The world changes because my light bulb can turn off at sunset." You're like, "Wait, what?" That's not a whole business.

Zach (40:37): Yeah. But that's interesting because, I think there's the consumer story.

Avidan (40:40): Totally.

Zach (40:41): I think there have been a lot of failed venture back businesses that started from a premise of we're building nest for X and nest ended up being a great exit, but most of those other businesses didn't do well, don't exist anymore, had small exits. You do have other cases where, Peloton for instance being an example. Now, Peloton's had all sorts of troubles in the sort of post public world. But as far as just being a venture backable business, I mean, they went and had a great IPO. I would expect that the folks who backed Peloton did well.

Avidan (41:19): Yes.

Zach (41:19): Is that a business model thing? Is it just that Peloton was a business with a subscription model and you consumer smart home product X that didn't make it or had a crappy acquisition, they didn't?

Avidan (41:31): Well, they, I mean, that's a perfect example. What is the value proposition? What are you replacing? Is there a willingness to spend? There is no world of cycling gym trainer where people don't pay on an ongoing basis.

Zach (41:48): [inaudible 00:41:48]

Avidan (41:48): Some pay monthly, some pay once per session. You had Flywheel. There was just so many things going on where there was a trend of, I take a group class, I pay 10 bucks a session or 20 bucks a session. I pay a hundred bucks a month or 200 bucks a month. Whatever it was. But there was something systemic about that. There was something about that setup that technology could unlock. Technology unlocked your ability to move to the suburbs and continue to do the thing you did when you lived in the city. When you were a young urbanite, you and your girlfriends could go spin together and it was a group experience, you had a trainer. Peloton really was just recreating that experience.

They needed to build great hardware to do it. They needed to build great hardware, but most importantly, they needed to build great content. The hardware was delivery of an experience that people were already paying for. It was a time and place that was perfectly done. It was not just about COVID. It was about the trend line of group fitness specifically around cycling ...

Zach (42:59): There was Soul Cycle.

Avidan (42:59): Right.

Zach (43:02): Right. You're taking an example where you're saying there was a business being displaced, Peloton was not displacing the at home bike. They were displacing the gym or the class. Those have ongoing revenue associated with them. You're not forcing a business model change. You're jus taking something from one place and putting it another, whereas, a more classic consumer product, if you're making a smart toaster ...

Avidan (43:31): That's exactly what I was thinking too.

Zach (43:34):

You're a coffee guy. How many times have you been pitched a smart coffee?

Avidan (43:37): Oh my God.

Zach (43:39): Actually, how many do you think?

Avidan (43:43): 40, maybe 60. Yes. Probably in the 60 range. That doesn't include the ones where I just got the pitch and wrote a very nice email to say, "Unfortunately, this is not [inaudible 00:44:05] to me."

Zach (44:04): If you include just the pitches, how many?

Avidan (44:08): You mean including the emails?

Zach (44:09): How many times has someone sent you an email and say, "I'm making a smart coffee maker."

Avidan (44:14): Yes. A hundred plus. A hundred plus easily. The reality is, in a lot of those situations, I think it comes down to the nest of X, then got replaced with Keurig of Y. People really telegraph the Keurig of X they're like, "Well, I'm just going to build the modern Keurig. We're going to build the Keurig of high quality coffee." At the end of the day, people then took this Keurig of something model, I mean Juicera was a great example and people weren't thinking about the value proposition of the connected device. Keurig itself made a bunch of missteps around what they believed the next generation of Keurig was.

For example, I don't know if you ever saw this device called the Keurig Kold. It's cold with a K, because Keurig. It was a Coca-Cola machine. They partnered with Coca-Cola and you had a pod, you'd press the button, it would chill the water, carbonate the water, and then pump it through the Coca-Cola. An undisclosed amount of time later, maybe six minutes later, you had a fresh Coca-Cola. When was the last time you said, "Wow. This Coca-Cola is so fresh." Think about the value proposition. Think about what you're displacing. The Coca-Cola in my refrigerator right now, that is probably canned a year and a half ago, when I crack it open it's fresh. In juice, yes, to a certain degree, you were like, "This juice is super fresh," but what's the difference between juice that was pressed 30 minutes ago, three minutes ago, or three days ago. We have technologically made those pretty fungible and like the consumer doesn't realize the difference.

I think it's about the value proposition. Making sure that there's a willingness to spend in a way that's ongoing, or that your technology is able to generate a true reason for being a service delivery mechanism, a platform for connectivity or something more than that. I think so many times people were just like, "I'm going to do this of X or I'm just going to stuff a chip in it. People will pay because well now it connects to wifi and therefore it's more valuable."

I think actually one thing that's even crazier is the wifi cellular thing. Speaking of getting pitches. I got pitched by a bunch of home security companies. For the longest time, I couldn't wrap my head around why a company like Simply Safe was doing so well. We didn't invest in any of these security companies because at the end of the day, we kept on talking to them and they kept on putting more and more chips in things. They were like, "It's got wifi. It's got Bluetooth. It's got cellular." We ended up doing a really interesting analysis of why did Simply Safe have an amazing attach rate to their recurring revenue and somebody who had wifi in their product did not? It was because the consumer knew that they had to pay for something when it had a cellular chip inside of it. But if it was wifi, they said, "Well, I'm bringing the internet to this thing. You're using my wifi."

The revelation we had was the attach rates for adding remote monitoring for Simply Safe, where it was a cellular only option ... They said, we will turn on an ATNT connection or whatever it was and it's going to be 10 bucks a month. That attach rate was significantly higher than the people who said, "Well, when you turn on our device, it has wifi and immediately is connected to the internet." There's a consumer expectation that they get certain services. If you walked up to somebody who's building IOT and said, "Hey, by putting less radios in your device, you will make more money." That's counterintuitive.

Zach (47:53): I mean, a lot of that comes down to a business's ability to activate a subscription model, at some point in its future. Which doesn't necessarily have to be out of the gate. Transitioning from not having a subscription model to having one is challenging. There's a bunch of ways to make that happen. There's a bunch of businesses where you just can't. If consumers look at a wifi connected product and say, "I don't feel like I'm supposed to be paying a monthly fee," then they won't, where if it's cellular, they might.

Regardless, if you come down to it, is it as simple as saying that the false positives are places where people thought that a product would be very successful, but it didn't have a subscription revenue associated with it. At the end of the day, that made it a good product, potentially, but not a venture backable product. That's where the core false positives have lived for the last decade.

Avidan (48:44): Those are the most visible false positives. Because you see the product in market, you see it out there. It is successful. People have bought it. People are using it.

Zach (48:51): People love it.

Avidan (48:51): It still went bankrupt, because they predicated their entire business model on continuing to raise venture capital dollars. The venture capitalists were like, " Well, where is your recurring revenue?" Because going back to that venture model, is you need recurring revenue because it's stacks. If you sell a million widgets, next year to double your revenue you have to sell 2 million widgets and then 4 million widgets. Whereas, if you get recurring revenue, you can count on the revenue. I think that recurring revenue is as simple as saying continue to provide value, ongoing value, to the customer means you can charge them for it.

I think a lot of times people just took for granted that, well, because it's connected, you're going to have this ongoing value or a dependency on ongoing support and somehow people would pay for it. I think that most companies didn't really wrap their head around how do I provide ongoing value to my customer? What ongoing value is unlocked by this connectivity? That is the holy grail. That is the holy grail. I think people have started to figure out things above and beyond just the willingness to pay to keep the connectivity going, and instead saying, "Okay. What is the ongoing value I could provide the customer? Or, what can I continue to sell the customer in consumables?" This is why everyone was obsessed with Keurig. Can I pull something off my connectivity to say, "Oh, by the way, I can sell you a subscription to supplies."

You see that with a lot of companies where there's a consumable attached to it. The reality is the consumer doesn't do a great job of keeping up to date on their consumables, or better yet, wants it to be turnkey. Just says like, "Hey, when there's a problem, just send me the thing that I need." Then, there are other people who are unlocking even wider opportunities where they say, "I'm going to put connectivity in. I'm going to sell better insurance policies. Or, I'm going to connect into people who are trading derivatives on how much grain is in a silo. Let me put a piece of hardware to like look into that silo." I think the most interesting versions of it are people who are truly transforming industries from top to ...

Who are truly transforming industries from top to bottom through hardware enabled transformation. So when you think about things like robotics, and I don't mean robots that are like R2-D2, C-3PO. I mean, people who have built fully autonomous connected systems that change the world and can react to changes in the world.

So we have companies like Tortuga, they build a robot that harvests strawberries. They're not just harvesting strawberries. They're also harvesting other fruits, but monitoring how much will this vine yield in the next three days, three weeks? Because I could see how many little green strawberries there are. Big green strawberries, half green, half red. That's not just the value of what the robot is doing when you initially decide to pay for it. Or other companies like Dusty Robotics that's painting layout on construction sites. They're able to take the digital world of your BIM model and lay it out on a construction site. But that unlocks all sorts of amazing opportunities where the people who come into the construction site are no longer interacting with the plans. They're interacting with what is painted on the floor, if it's a QR code or information about which subcontractor on what day is doing work.

I think that's the next step of building connected hardware is thinking about not just how do I charge somebody to keep the lights on? But how do I transform the way they do business because I've brought intelligence to the edge.

Zach (52:40): Yeah. Yeah. What's your advice... Actually, hold on. Before I get to that question, false negatives. So what are the areas where you've seen these are the companies, people are not investing in these, they should be and there's something about this kind of business that makes it really hard to raise venture capital for, even though it should be... This should be venture backable. It has all this stuff that leads to good returns for a PC fund.

Avidan (53:08): Yeah. That's a great question. Where are all the misses? I think this is probably a dig on the entire venture community, but people really have a hard time imagining industries they're unfamiliar with. The reason why there's so much enterprise SaaS, and networking, and security, and social media investments being done is because when you look at the history of so many venture capitalists, it's where they worked. It's where they were product managers and project managers and finance people and sales people. Those are the industries they know.

The industries that are getting missed are the industries that don't run in California on the 101 Freeway between San Francisco and San Jose. If you drive that freeway, you will see the industries that are getting funded. The amazing thing about connectivity, and connected devices and IoT and hardware is that they're oftentimes disrupting offline industries. Those offline industries are massive. It's agriculture, it's mining, it's construction, it's waste management. It's all of these huge industries and people will come up with all sorts of reasons why they don't think that industry is venture backable.

The funniest is when they say well, the TAM isn't big enough on that and you just sort of... It's hard. You can't undo the way they think about an industry, but then when you tell them it's a multi-trillion dollar industry, they kind of can't wrap their head around it because they can't empathize with the problems of the people in that industry and how transformational this solution can be. So you end up in a world where there's a very small audience. Now, there are a bunch of VCs who have found a way to reach into networks of people to learn about the industries they're not familiar with. But at the end of the day, it's a small group and the Venn diagram overlap of people who are willing to invest in hardware or deep tech companies and willing to look at those big out of sight industries is very small. I think that there are a lot of misses happening.

Zach (55:24): Yeah. That makes sense. If you're selling into agriculture, in order to understand whether a farm would buy this product, you have to be able to empathize with the farm, which you might not be able to. The farm and the farmers and the people who own that business, which you might not be able to do because you've never done any business with them before. So you can't figure out whether this product has product market fit or is going to have product market fit.

Avidan (55:47): Calling your buddies in Napa to find out about their hobby vineyard, that's not calling a farmer. You're like I got a place in Napa and I'm going to call a friend of mine 'cause he's next door selling $200 bottles of cab. You're like, that's not American farming. So it skews people's perceptions of these markets.

Zach (56:07): Yeah. Yeah. That makes sense. Okay. So now wrapping up advice. If you are a founder who's thinking about creating a business that has hardware and software, you're thinking about... Let's just assume, I mean, aside from the fact that there's a lot of good businesses that aren't venture backable. If you are trying to build a business that is venture backable, or you believe you have one and you want to engage with the venture capital community and you haven't done so yet, or you're just trying to figure out how to. What's your advice to a founder to first A, guide their business towards being a good fit with this kind of capital and B, how to actually engage and talk to investors about these kinds of businesses?

Avidan (56:53): It's always hard to give advice because at the end of the day, each entrepreneur is taking a different path. But what I would say is that first question is a great first question. Do I actually want to build a venture backable business? Or do I just want venture dollars or do I just want venture validation? Do all my friends here in Silicon Valley or in New York or wherever you are, does it make you feel like your company is worthy?

I think that's a bad reason to go down the venture capital route. If you feel like you will not be able to sleep at night if the problem is not solved and it requires significant dollars and venture capital is the solution, I think there are a couple steps you should think through about. First understanding what about your business becomes venture appealing? I think that that is allow yourself the ability to think of the biggest version of the business.

Think about everything going right. Think about if I get this, and then I do this and then I sell to these people. Then question all of it and say, okay, how much of that is plausible? Can I really sell this far? Would people pay this much? How do I go to confirm that conclusion I've already made? I think you spend time getting out there and talking to people who are in the industry who would buy. Now, you can't take them at face value because a lot of people will say yes to a problem being solved or can't imagine a solution to their problem because it's self validating. It validates that they are important in their job or that they would say yes to anything that makes their life easier.

But you just have to really ask that question of what is the biggest possible vision of this business? How big can it become and do I see myself working insane hours and literally making my life dedicated to that vision of the world I want to see? Do I care so much about that end that I'm willing to take all these risks to get there?

Then in terms of reaching out to VCs, I would just say, it's don't be impatient. People in venture are trying to make decisions with very minimal amounts of information and they take a lot of credence into who makes the intro. So you'll have to work your way in to getting introduced to someone who might be one layer away from that VC and use that opportunity to get advice on how to hone your pitch. Because as that pitch is getting better and you've built a relationship, then the person has an incentive to then introduce you.

The warm introduction is so much better, but also the process you went through in order to get the introduction is making your pitch that much better because to be really blunt about it, you get one chance to make that first impression and talk about your business. Don't just DM the full pitch. Yes, I'll read it but you didn't get the opportunity to talk to me about it. You didn't get the opportunity for someone to warm it up and give you their insight and their perspective because the person making an introduction might say, "Hey, this is a really smart person. The idea's almost there, but I really think this industry's great because my cousin works in it."

You get more information and I think that that's what VCs are looking for at the earliest stages. So yeah, I'll read your DM. If you tweet at me, I'll read it. If you send me a cold email, I'll read it. But your odds of being able to get a conversation going will improve when you've spent time starting conversations with other people and it becomes a group conversation. 'Cause even after we've taken a pitch, what we immediately lean into is who else can we speak to about this company, this problem? Oftentimes it is the person that you chose to pitch first who then is making that warm introduction. I know people don't love that and they would love to just be able to email directly when they can, but odds improve.

Zach (01:01:17): So if you were to get a pitch for a smart coffee maker and it came in through a warm introduction versus a cold email, how much more likely are you to get on the phone with that person?

Avidan (01:01:28): Oh, infinitely. I mean, it's night and day because the person who sent the intro has spent some social capital. They've spent some social capital saying hey, I know you always read my emails. You always respond to my emails and I care about your time and I think this is worth looking at. If I get a warm email intro, I almost always, there are occasionally times when we say hey, this is too competitive, possibly close to another company we're looking at. But warm introductions from people we respect are significantly higher.

Avidan (01:02:15): We just did two deals where they came in through cold inbound. So it's not to say that we don't do cold inbound. There's plenty of deals that come in that we end up investing in via cold inbound. I would just say that there's a far larger number that don't get done and your odds of getting that first meeting are really about getting to know people who honestly forget about improving the odds of getting to meet with me. It's a great excuse to get on the phone with somebody and try to hone your pitch and think through the business model.

Avidan (01:02:49): The whole way Silicon Valley works is a pay it forward attitude of everybody willing to get on the phone and share whatever they think. By the way, if you just want to get on the phone with me to get advice, that's easy. If you're not asking for money, it's a much easier and that's how everyone in Silicon Valley is. Honestly, if you saw me at a cocktail party or you saw me at a coffee shop and you wanted to come up to me and chat for a couple minutes and hear my thoughts, I'm not shy. I will tell you what I'm thinking and that's great. But in order to get the full pitch and to lean in to do the diligence, it is a group effort. It's not just, if I can only get a conversation with that one person, I can get money. It's not how it works. It really is a lot. There are a lot of people involved.

Zach (01:03:32): Cool. Well, thank you Avidan for joining us today. Hope this has been helpful and if you have any other questions for Avidan or for myself, shoot us a note, reply to this in atoms and bits with the newsletter and yeah, thanks for joining us.

Avidan (01:03:45): Yeah. If you want, I'm @Avidan on Twitter.

Zach (01:03:49): Can you spell Avidan for those who...

Avidan (01:03:50): Oh, Bond. James Bond. I'm just kidding. It's A-V-I-D-A-N.

Zach (01:03:58): Also Root Ventures can be found at Root.vc.

Avidan (01:04:01): Root.vc. Have fun on the website.

Zach (01:04:03): If you like their website, then you're going to like Root Ventures. If you don't like their website, then you're not going to like Root Ventures.

Avidan (01:04:08): That's very fair. It is the ultimate... We actually thought about saying that. Who do you invest in? And we said, people who like our website. Root.vc, Have fun.

Zach (01:04:19): All right. Thanks folks.

Internally, Particle had… similar feelings. Some choice quotes from our Slack thread on the topic:

• “Would 100% ensure I didn't buy a BMW”

• “Very excited to live in the new microtransaction hellscape.”

• “Reminds me of Spirit Airlines ‘racing towards’ charging for carry-ons.”

• “I hate this with every fiber of my being”

But wait a second. Particle is an IoT platform whose customers ship hardware/software products, generally with subscriptions attached. This newsletter is only on its fourth issue and I’ve already advocated for “hardware-as-a-service” business models at least twice. So why all the shade cast towards BMW?

I believe that manufacturing businesses are going through a business model transformation that will lead to any high-value product ($10K+) being sold either partially or completely under a subscription model. Businesses that drive that transformation will become industry leaders, and others will be left behind.

But I also believe there is a right way and a wrong way to go through that transformation. And BMW’s approach is, frankly, wrong.

So, let’s use this as a case study. What is the wrong way to implement subscriptions for a hardware product? And what does that imply about what the right way is?

BMW’s Subscription Model

First, let’s get past the headlines and lay out the facts about BMW’s subscription model:

In 2020, BMW announced that they would begin selling features of its cars through “digital after-sales”. Here’s the relevant text from the announcement:

BMW already offers its customers digital services and additional vehicle functions in the form of digital after-sales, some of which are deeply embedded in the vehicle's software. In the US, BMW Drive Recorder is offered through a pilot program. In the near future, additional functions will be added that can access the vehicle's existing hardware and software, such as certain comfort functions or driver assistance systems.

With the option of subsequently booking additional vehicle functions via the ConnectedDrive Store, BMW is strengthening selection and personalization for customers, offering them maximum flexibility. BMW provides the hardware and software in the vehicle at the time of manufacture for the implementation so that it can be adapted later on as required and in accordance with customer preferences.  If, for example, vehicle functions were not yet requested at the time of purchase, they can be added later. A second owner thus can configure the used vehicle according to their own wishes.

Side note: I find it fascinating that they seem to be avoiding the word “purchase”, like they don’t want to admit that you’re going to have to pay for this stuff. You can “book” additional vehicle functions, they can be “added later”. Look folks, if you’re embarrassed to admit your own business model, you’re probably doing it wrong.

As promised, the BMW ConnectedDrive is filled with “digital services” you can buy (sorry, “add”). As of July 19, 2022, you can “add” a total of 13 features:

• Active Cruise Control with Stop & Go function (Purchase)

• Adaptive M Suspension (Purchase)

• Apple CarPlay Preparation (Purchase)

• BMW Drive Recorder (Purchase)

• BMW Safety Camera Information (Subscription)

• Driving Assistant Plus (Subscription)

• Front Seat Heating (Subscription)

• High Beam Assistant (Subscription)

• IconicSounds Sport (Purchase)

• Map Update Package (Purchase)

• Online Entertainment Voucher (Subscription)

• Service Inclusive (Warranty-style purchase, limited to 5yrs/100K km)

• Steering Wheel Heating (Subscription)

Putting aside “Service Inclusive”, which seems like a standard AppleCare+ style model of expanding and extending the warranty, six of these upgrades are one-time purchases and six of these upgrades are subscriptions.

If you’re like me, you’re probably reacting to this list as follows:

• It seems reasonable to charge a one-time fee for upgrades that I would traditionally purchase as part of a premium trim package (like “Adaptive M Suspension”). Rather than buying a basket of features up front when I buy the car, I can select only what I want and buy it whenever I feel like it. I’m good with that.

• That said, the longer this list gets, the more I feel like I’m being nickeled-and-dimed. “Bundling” does have a clear benefit to the customer: avoiding “microtransaction hell”.

• I kinda love that you can pay 99 euros to play the BMW engine sound inside the vehicle. That sort of reminds me of buying ringtones circa 2004.

• Once we hit the subscriptions, things go off the rails. The seat heaters are the feature getting all the flack, but “high beam assistant” and “steering wheel heating” feel equally wrong.

This subjective, intuitive reaction to the list of 13 features that BMW offers as “digital after-sales” shows that some fee structures seem reasonable and others don’t. As we build a framework for the right and wrong ways to add subscriptions to a hardware product, let’s pull in a counter-example from the automotive world – Tesla’s Autopilot.

Tesla Autopilot

If I go onto Tesla’s website to buy a Model 3, I am presented with a slick user interface to customize by car by adding/removing features. At the moment, I can spend as little as $46,990 or as much as $77,990 on a Model 3.

That $77,990 price at the top end includes a whopping $12,000 for “Full Self-Driving Capability”. Tesla has gotten a lot of shit for how they’ve marketed this feature by overpromising and under-delivering on delivering a true “self-driving car”, but to give credit where it’s due, if you put aside the cost and the marketing for a second, this feature is pretty damn nice. Tesla has been a market leader in driver assistance features since it originally launched Autopilot in October 2015, even if other automotive manufacturers have since caught up.

But Tesla didn’t just lead the pack on driver assistance tech. They also innovated on the business model. Because if I choose not to pay $12K for the feature up front, I can choose to activate a “Full Self-Driving Capability Subscription” later.

There is something inherently less icky about Tesla’s subscription model here. There are some major differences between BMW and Tesla’s subscription models that influence how customers will react to them:

• Tesla offers one subscription instead of many.

• Tesla’s subscription makes a very expensive feature more accessible.

• Tesla emphasizes the ability to unsubscribe at any time and thus only pay for FSD when you’re actively using it.

• Tesla uses its subscription to facilitate evaluation; customers who are unsure whether they’d like FSD can sign up for a month and try it out. This is important because a) most people haven’t tried one of these driver assistance systems when they buy a Tesla so they might be unsure of whether they’ll like it, and b) the feature keeps improving so if you try it and don’t like it you can give it another shot after a year or two of improvements.

• And now the big one: Tesla’s subscription is applied to a feature whose value is delivered through software rather than through hardware.

This last bullet is, in my opinion, the most important one.

As the electronics products that we own have become more intelligent, we’ve become trained to separate “hardware value” from “software value”. We attribute features that rely heavily on physical characteristics of the product – like heated seats – to hardware value that we expect to purchase upfront. In contrast, we understand that some features require a heavier software investment; those features have user interfaces presented on screens and get downloaded over-the-air and are improved over time by the manufacturer. We understand that these features require an ongoing cost to deliver, and therefore we are more willing to entertain the idea that the manufacturer has earned an ongoing fee.

A Better Subscription Model for BMW

Given the nature of the features that BMW is attempting to provide via subscription, is there a different model they could pursue that wouldn’t piss people off?

Yes! At least I think so. If I were BMW, I would bundle the relevant features into a “Winter Driving Package” that would look something like this:

• Heated seats

• Heated steering wheel

• Remote engine start, windshield defrost, and climate control (heat up the car and notify you via push notification when it’s nice and cozy inside)

• Advanced driving features to facilitate safer driving on ice and in snow (look, I don’t know what these features would be exactly, but I’m sure the smart people at BMW could figure it out)

• $49/mo subscription, activate/deactivate at any time (encouraging customers to activate for the winter and deactivate in the spring when the weather warms up)

• Marketing campaigns in the fall to sell a discounted “full season” winter driving package (e.g. $249 for access to the Winter Driving Package from October 1 to April 31)

Why does this feel better than BMW’s current approach? Because it’s using a subscription model to do something that provides an actual benefit to customers. Not everybody lives in a climate that requires a winter driving package, not everyone is willing to pay for these features, and nobody needs them at all during the summer. If we design subscription packages around how customers want to use our products, we can build subscription revenue in ways that make customers happy.

Audio Only 👇

This interview is about an hour long, and is best consumed in video or audio. That said, if you’re looking for the Cliffs Notes, read on, and further down we’ve included a transcript of the interview.

Atoms & Bits will alternate between essays and video interviews like this one; tell us what you like and dislike as we’d love the feedback for future issues and episodes.

Key Takeaways:

• The Rise and Fall of Boosted: Sanjay and his team created a high-demand product (the Boosted Board, a high-quality electric longboard) that had a successful launch on Kickstarter and sold well. Boosted was a solid business, but one whose business model (selling skateboards at a margin) wasn’t a great fit for venture capital investors, who drove the company to grow quickly. Ultimately, Boosted fell as a result of tripping a covenant from a debtor whose money they took to fuel growth, which led to a downward spiral and eventual collapse of the business.

• Skip: Sanjay dives into how micromobility developed into an industry whose winners and losers come down to how well companies in the space partner with regulators; it’s created some interesting dynamics very unique to micromobility, which continues to be an exciting but very challenging industry.

• Overall perspectives on HW/SW businesses: Selling widgets under a traditional hardware purchase business model is not necessarily the best way to create a high-growth business.

Transcript:

Intro

Sanjay (00:00): So, we met Tony Hawk. And we went to X Games. And we were trying to meet all these people who were in action sports. And right under our noses, the reason Casey loved it is because it was like an ultimate filmmaking tool.

-

Zach (00:14): Hi, everyone. My name is Zach Supalla. And this is Atoms and Bits. We're here today talking to Sanjay Dastoor who is the founder of Skip and Boosted. Sanjay, thanks for joining us.

Sanjay (00:22): Thanks for having me.

Zach (00:23): Well, you are our first guest on this video podcasting that we're doing. So, we're going to be figuring a lot of this as we go. But to start, I'd love to just give you the chance to tell us your story. Who are you? What have you done? What got you into this stuff in the first place before we start picking up all the details?

Sanjay’s Backstory

Sanjay (00:43): Yeah, sounds good. I'd say I started working on startup stuff accidentally in hardware. So, my background, I grew up in the South, went to college in mechanical engineering, electric engineering, wanted to do robotics work basically from starting late high school. So, I took a lot of classes in between embedded systems and EE and ME and went to grad school.

I got to intern at some cool places like SRI and at JPL done in LA. And then accidentally ended up working on a project that turned into a startup and have been working on startup-related hardware projects ever since.

Zach (01:17): Okay. Was that Boosted?

Sanjay (01:18): Yeah. So, Boosted was the kind of accidental project that turned into a company.

Zach (01:22): This is while you were in your PhD program, is that right?

Sanjay (01:24): Yeah, two of us were still working on our PhDs. And one of us had finished our master's and was working. And we were doing this on the side until we raised our first little bit of funding and went through YC. And then it finally became like a company within full-time founders. And that was where that really started.

Zach (01:42): Okay, interesting. So, before Boosted was a Kickstarter campaign, there was a project. Tell us a little bit about what did it start as? What were you first working on?

Sanjay (01:50): Yeah. So, there were two parallel projects. So, my c founder, Matt, have had this idea. So, we'd become friends in grad school, snowboarding and riding motorcycles. And he lived in San Francisco. I was still living near campus. But he was living in San Francisco and wanted to basically combine those two things as a way to get around.

So, he was envisioning a snowboard for the road that you could get from one place to another with and needed some help on commercializing it. And we were friends and he was like, "Hey, can you help with some of the circuits and some of the code?" I said, "Sure."

And so, we worked on it together. In the meantime, one of my labmates, John was, like many of us, we were going back and forth between buildings on campus especially to check on machines or check on processes and see, does this thing dry yet or is this thing done being cut yet or something? And right then, we were seeing these motors and batteries that then enable other industries like drones and e-bikes first become commercially available.

And John was using a longboard. He's like, "Man, if I could add this to my longboard, it would be amazing." And so, he was solving a problem he had to get between buildings on campus very quickly and not have to spend 40 minutes round trip walking to this other building and back. And Matt and I were working on this side project with a similar goal but designed for Matt's neighborhood environment in San Francisco.

Zach (03:11): So, I'm trying to picture what is a snowboard for the road look like? What was the ...

Sanjay (03:15): So, Matt's prototype that he had in mind was actually ill-conceived. I mean, it was a good idea, but the implementation that we had was really that which was designed around a freeboard. So, it was a skateboard with two casters on it and so you could actually pivot from one edge to the other like a snowboard. So, super dangerous, really hard to ride, incredibly hard to ride.

But the idea was basically, we would just carve back and forth a board like a longboard. And so, a Boosted board does feel like that when you're riding it. It feels like you're carving on a snowboard. But that was the original idea was Matt had this idea for how he wanted to get around, wasn't sure about the architecture. And then John had his architecture figured out, but was thinking very much about like a small use case or on campus and Matt was thinking much bigger. And so, the three of us ended up working together.

Zach (04:04): Got it. So now, how did you go from working on that as a project to YC Kickstarter ... Which of those happened first?

Sanjay (04:13): So, we started working on together I'd say around the summer of 2011. We did YC in the summer of 2012. We did Kickstarter in September of 2012 so right after YC. So, for the first year, we were just working on it as a side project. We were all still either in school or working. This was just evenings and weekends thing. We just put a little bit of our own money.

And we didn't expect it to be like a startup or a venture-backed startup, for sure. It was really just something fun we were trying to build, maybe sell a few of them. That was the idea. So, we looked at how do we file a patent or how do we work with somebody who is willing to work with our low budget on the industrial design or on graphic design for some pieces of it?

So, those were the early challenges that we had. We're really just doing this as broke students. And then only once we got into YC and we had $100k in the bank for the first time which was this incredible amount of money, then we said, "Okay, we can go on this full time. We can still pay our rent. We can try this out for three months and see how it goes." And then we just kept going.

Zach (05:12): When you applied to YC, how similar was the thing that you were applying with to the product that you ended up actually bringing to market?

Sanjay (05:19): It was pretty similar. I'd say the biggest thing we learned in YC from testing between the beginning when we applied and when we shipped it was that people cared about reliability. There were two types of users. There were people who viewed it as a toy. There's people who viewed it as a vehicle. And the toy people were willing for it to break all the time, but they didn't really care if it broke. They didn't care the way the vehicle people cared. The vehicle people said, "I need this to work every day."

And so, that became a framework we thought about later in making decisions like are we designing for the nice-to-have or the need-to-have type of user or a feature. We were always thinking in terms of those two groups. So, what we applied with was not reliable. It would break constantly. We built a few and tested those. And that really reinforced for us that the people who wanted to work, we had to do a lot of engineering to go from a prototype to actually a reliable everyday vehicle.

Zach (06:09): Got it. So, when you eventually launched the product on Kickstarter, I was looking at it earlier, the framing you had around how you described the Boosted Board was really interesting because you didn't describe it as an electric skateboard.

Sanjay (06:24): Yes, that's correct.

Zach (06:26): If I'm remembering correctly, it's the world's lightest electric vehicle.

Sanjay (06:29): Yes. We never mentioned the word skateboard on Kickstarter.

Zach (06:33): Right. It's definitely a skateboard.

Sanjay (06:35): It is.

Zach (06:36): Talk about why you framed it that way.

Sanjay (06:39): So, we actually learned that lesson from Eric at Pebble. So, when we had applied to YC, there weren't that many hardware companies that had gone through YC. And Eric was one of the few with Pebble. And so, we talked to him and wanted his advice just on launching Kickstarter. They've done a really successful campaign with Pebble.

And he said, "You could think about the people who you're speaking to like your audience on Kickstarter in terms of concentric circles. And you've got this inner circle of folks, the smallest circle of people who will immediately get it and they will immediately understand how this thing works. But then there's another circle around that that's a bigger group of people but that it won't be immediately apparent how they would use it."

And he talked about on the Pebble video how they showed the way people would use it and not just what the features were, they would show use this in a sports situation, use this for meeting. In the early, early days, they invented the smartwatch, what are all the different ways you could use a smartwatch. You could track fitness. You could track reminders. You could use it for reading texts on your phone. You can do all sorts of things.

So, we thought about it in those terms of everyone's going to know it's a skateboard. And so, people who really love skateboarding or longboarding will already get it. But I had never skateboard him. And so, for me to even consider this as something I would use, I wouldn't be like, "Oh, a skateboard and it's electric, well, that makes me want it more." I would have to think about like, "Oh, I can use it to get on the bus. I could use it to ride around my neighborhood. I could use it to get to class."

And so, the way that you would use it became how we framed the narrative on Kickstarter as opposed to just like what the product physically was.

Zach (08:14): Right. It's funny. I remember when it went live, falling very much in love with that product, also never having skateboarded.

Sanjay (08:21): Yeah.

Zach (08:21): And I think that framing helped because I wouldn't necessarily see myself riding a skateboard, but I can see myself getting around on this thing.

Sanjay (08:30): Exactly.

Zach (08:30): And so, it makes sense.

Sanjay (08:33): Yeah. And we heard that from a lot of people that some people said, "I never longboarded before but this made me want to learn," or, "I could envision myself." So, we've learned later, GoPro people call this I think the moment of inception which was people wake up in the morning and think, "I want a cup of coffee."

And so, if you're starting a coffee shop, then you're already dealing with a need but people didn't wake up going, "I need an action camera." So, how did you create the sense for someone learning about GoPro that they would use this and it would be part of how they snowboarded or went scuba diving or whatever they use it for? And so, we tried to create that. And it turns out, yeah, a lot of people who didn't skateboarded or longboarded still felt this was appealing.

Zach (09:11): Yeah. So, you did about a half-a-million dollars of sales via Kickstarter campaign.

Sanjay (09:16): That's right.

Zach (09:16): How did that compare to what you thought was going to happen?

Sanjay (09:18): So, we actually designed it that way. We wanted to make sure that we didn't sell too many. And that actually was a lesson we got from Eric at Pebble again which was if you are building 100 of something in hardware and all of a sudden you have to make 1,000, most of your processes will break. But if you set up your process for 1,000 and you only make 100, then you're paying a lot more for overhead or it's just a lot of extra work.

So, we said, "Okay, well, if this is successful on Kickstarter, we want to create an incentive for people to get up to the first few 100 units, but we don't end up with 1,000 and then take a much longer period of time and actually need a lot more cash to buy the parts." And that just becomes a huge issue to try to scale up your production systems ahead of time.

So, we designed the tiers in terms of the discounts or when you could get it earlier or when you would get the boards where there's a decreasing advantage to doing Kickstarter versus just ordering it after Kickstarter was over. And so, we probably could have gotten more if we'd offered like a deep discount earlier but we purposely didn't.

Zach (10:24): So, how many did you end up selling? How many boards?

Sanjay (10:27): It was around 350.

Zach (10:28): Okay. That's not a crazy number.

Sanjay (10:29): No, it's not crazy. But I mean, to us, it was. We only built five. That's a lot. And we were committing to not just sell 350 but also take care of the people who had bought them after they bought them. And that was one of our big lessons during YC was we sold five and realized there was at least as much work if not more to take care of them after they bought it than it was to make the devices in the first place.

Zach (10:49): How much do you think it costs you to design and manufacture? If you accounted for the whole cost of those first 350 units and support and so on, how much did that cost versus your half-million dollars of revenue?

Sanjay (11:05): So, it was probably about $2 million to go fully through to production and start supporting the people after Kickstarter.

Zach (11:21): There's no world in which ... If you imagine a world without venture capital, is there an alternate universe you could have created where that product could have existed in some similar form without venture capital or was it just not possible?

Sanjay (11:36): I don't think it would have been possible with the engineering work that we did. I think there are types of products like, let's say, soft goods — like Peak Design is another really popular set of products on Kickstarter where they're machined metal pieces or they're backpacks, it's a lot of camera equipment. But there's no circuit board that can blow up. There's no battery. There's no motor. There's not a safety critical product in that sense.

Sanjay (11:59): And so, you don't need to do a lot of testing in the expensive way you would need to. It's something that you're going to get on and go 15 miles an hour on the road with. And we had to design certain things like regen braking where we had to push a lot of current back into the battery when you slow down. And those things didn't exist.

Sanjay (12:18): So, I think there is a way to do it without VC and there are electric skateboard companies out there that do this, but they have to be very disciplined about what challenges they take on. Whereas we said, "Okay, this is the product vision we have. And in order to realize it, we have to spend this much to get there."

Zach (12:33): When did you raise money for the first time above and beyond the sort of Y Combinator?

Sanjay (12:39): Right after YC. It's at the end of YC Demo Day. And then probably for another six months or so after that, we had a rolling close of at that point convertible notes (basically the SAFE way).

Zach (12:53): So, you went into the Kickstarter Campaign already having some financial backing so that that was like theoretically possible to accomplish?

Sanjay (13:00): Yeah, we didn't know it would cost us that much to develop it. And frankly, if we had not raised money and adjusted the Kickstarter money, we probably would have been exactly like a lot of other Kickstarter projects that failed to deliver or that deliver something way below what people are expecting.

Zach (13:12): Yeah. So, it costs you $2 million to do in the end. What would you have guessed? If I'd asked you the day your Kickstarter campaign closed and you had half-a-million dollars in revenue, how much this is going to cost to make? What would you have guessed at that point?

Sanjay (13:26): Probably around half-a-million to maybe a million.

Zach (13:29): Okay. So, you need like a little bit extra cash but not a crazy amount of extra cash.

Sanjay (13:33): Exactly. And then we learned, "Oh, there's this limitation with a subsystem," or, "We need to certify this thing for FCC and that's going to cost $15k." We never accounted for all those little things.

Zach (13:44): Was there any one ... Is it just a bunch of little things or is there one thing you're like, "And we just totally underestimated this one problem?"

Sanjay (13:52): Yeah, the battery was a big thing that we underestimated the cost of. Batteries, the reason Tesla's work in terms of the battery pack because you're only getting a little bit of energy from each cell but you have thousands of cells. The Boosted Board is different because you still need a fair bit of energy, not as much as a car, but you're getting potentially hundreds of watts of power from each cell. And that's far beyond the standard capacity of a standard let's say Tesla cell.

Sanjay (14:18): So, Tesla couldn't make a car that only went 20 miles using those cells. They need a lot of cells in parallel to get enough power to accelerate the car. So, range and power go together. Because this was a super compact battery, we needed a lot of power per cell. So, that limited our options. We needed to do regen braking back into the pack. And we wanted the pack to be really safe and we were very conservative about battery pack safety.

Sanjay (14:42): And this is pre-hoverboard. There were no standards out there that ... There were no rules about what you could and couldn't ship really. The further along we got with the battery process, the more we realized we have to take a lot of that in-house or do a lot of extra work to test it for us to feel comfortable shipping it. And so, the R&D cost was higher. The timeline to ship it was longer. And the per-pack cost, the cost of actually making the pack was much higher than we expected.

Zach (15:07): How delayed off of your initial shipping timeframe were you?

Sanjay (15:12): If I had set the shipping timeframe, we would have been really delayed. John, my co-founder had been much more concerned. He's like, "No, we have to tell people." I think it was maybe 12 months. And I think we shipped in 15. So, we were close.

Zach (15:26): That's not bad as far as Kickstarter Campaign.

Sanjay (15:28): It's better than most campaigns. But we took on a lot. I mean, if we had known, we probably ... We learned a lot about how hard it was by doing it.

Scaling Boosted

Zach (15:39): Right. So, Boosted Boards went on to be ... I mean, sometimes you created an industry that didn't exist before or a category I suppose-

Sanjay (15:49): Yeah, we were one of the early ones to use modern components in that category which opened it up a lot.

Zach (15:54): At the end of the day, about how many boards do you think ... Cumulatively, how many Boosted Boards are out there?

Sanjay (16:01): Probably over 100,000.

Zach (16:02): And what did the shape of that look like? Was it a sort of a growth curve that sort of looks the way that, I don't know, you expect startups to grow? Or is it-

Sanjay (16:14): At the beginning, yeah. Our first year, we sold around 2,000 boards. So, we had the 350 from Kickstarter. I think we finished those up ... We started shipping them around January of 14th. We were building five a week. That's the pace we started at. We finished shipping those in the first quarter of the year so by the end of March. And then, we ramped up production through the year. And we were doing all the final assembly in Mountain View, locally.

Sanjay (16:43): So, by the end that year, we did about 2,000. The next year, we did about 6,000 or 7,000 units. And then the next year, we ended up doing around 8,000 or 9,000 but because we have stopped shipping our second-gen product because of a battery recall that we did, we probably had demand for about 20,000 or so. And then, the year after in '17, I think we did about 25,000 or so that year.

Zach (17:10): Was that organic? Was that because of marketing? What was driving the increase?

Sanjay (17:14): Yeah, we tried everything. We tried digital ads. Initially, we were direct to consumer and then we went into stores like BestBuy and skate shops and Amazon once we could afford to give them some of the margin on the product. We worked with influencers, Casey Neistat, probably the most famous, although that wasn't really an intentional marketing choice where he just really liked the product.

Zach (17:37): He really liked the product.

Sanjay (17:37): And a lot of it was word of mouth. A lot of it was test rides and really happy riders telling their friends and their friends trying it and then buying one as a result.

Zach (17:50): Yeah. I think the first time I rode one was you guys had a test ride that was in Hayes Valley. And there were folks that were going around that little area. And that was the first time I hopped on one. And I was about to tell you I was floored which I literally -

Sanjay (18:04): Did you actually fall?

Zach (18:06): I didn't fall on my ass, but yeah, I fell. I stepped off the thing.

Sanjay (18:10): Yeah, you stepped off. That's totally fine.

Zach (18:13): It was a blast, yeah. Incredible amount of work. So, you talked about your margins improving and being able to bring on channel. What margins did you guys have on a product once you were operating at scale?

Sanjay (18:24): Once we operate at scale, so we did three generations of the board. I was there for the first two. The first generation ones, we really struggled to get above when we included blended margins for like channel sales. We struggled to get out of the 10-20% range which is really too low. You need to be higher than that.

Sanjay (18:44): Direct sale margins were great. But we found that especially Amazon and having a somewhat trusted set of reviews of third parties, we weren't influencing reviews and we couldn't edit which reviews showed up. So, the fact that I think we had 90% five-star ratings on Amazon, that was a huge driver for us. And for us was, in a sense, selling an Amazon was much better marketing spend than buying an ad because the value of those reviews was so high in the purchasing process.

Zach (19:09): What margins were you guys taking when you did a direct sale?

Sanjay (19:13): More like mid-30s, mid to high-30s. So, we were giving channels anywhere between 15% and all up to 25%.

Zach (19:21): Right. Okay. Yeah. And that was enough because I know that the margin expectations generally of brick and mortar are usually higher than digital Amazon, things like that. Is that enough to get brick-and-mortar folks selling your products?

Sanjay (19:32): Yeah. So, what was interesting for us was that our price point was so high. So, the average skate shop was selling $100 item at most or even like $5 items on stickers or skate wheels or something. And so, their expectations for margin on those items were much higher. Whereas you wouldn't pay for ... A car dealer doesn't take 30% margins on a car. So, the higher the price the item is and especially the more it's moving, the more they're willing to tolerate lower margins for themselves.

Zach (19:59): Got it. So, as you improve margins, where did it sort of max out? What was the best?

Sanjay (20:06): The best we got blended was probably around 20 or low-20s. That was going to get fixed with a third-gen. And actually, we learned much later, and I think Tony Fadell talks about this as well with Nest, where you want to be deliberate about generations of your product and not just think of your product as shipping like a single thing but choosing what to validate with each generation.

Sanjay (20:31): And so, in the first generation of let's say an iPod, when he talks about this or I think the Nest thermostat, you're really validating that people want the product but you're optimizing around the product and not around the margin. And then, you can start to optimize for those things in Gen 2 and Gen 3. And I think his rule is like three generations to really dial everything in.

Zach (20:48): Right. So, that was sort of on the road, the third-generation was where those economics ... That was after your time at Boosted, is that right?

Sanjay (20:54): Exactly, yeah. But the target for those when I was there was that we wanted to get to blended mid 30s to high 30s.

The End of Boosted

Zach (21:00): And that was where the business would have felt like comfortable like this is a good margin. You can put money back into the business. You can feel growth. So…now, let's talk about the end times at Boosted which was after your tenure there. But tell us, from your perspective, what happened at the end? What drove the demise of Boosted? And what are the lessons that you take away from it?

Sanjay (21:24): So, when I left, we had done about $29 million revenue that year in 2017. We hired a new CEO. He had come in with really good experience out of larger tech companies, but he'd also come out of the same robotics doctoral program that we had left to start Boosted. And so, he had good context for what the problem was and for how we thought about things.

Sanjay (21:51): The biggest problem I would say and it's hard to tell because also, I wasn't there. I was involved on the board for a period of time, so I don't have all the details. But one of the things that helped us a lot was we always had trouble building enough boards. And so, we never were making huge inventory commitments in the first few years of selling the product. And that turns out to be an easy thing that kills companies when they have a physical product is they have cash but it's not cash, it's actually inventory.

Sanjay (22:20): And they can turn that into cash. But if let's say they build a bunch of units in anticipation of holiday sales that don't happen, not only did you not get the revenue from those sales, you're also sitting on inventory that then messes up your supply chain commitments in the first quarter of the next year. And sometimes you can even get stuff returned.

Sanjay (22:39): So, if you work with a bigger like a big-box Best Buy-type of company and they don't sell all the stuff they got for Christmas, sometimes they have the right to give that back to you and you can get a refund.

Zach (22:52): Pebble had this challenge, right?

Sanjay (22:53): Exactly, yeah. This is what happened with Pebble I think. So, in the first few years, we always were ending the year with almost no inventory. And then, after I left, I think we got more aggressive as a company about the expectations we were setting with investors for growth. And continuing that trajectory that I talked about earlier, we got more aggressive about maturing our supply chain which we had to at that scale, but maturing the supply chain then working with partners who expected larger MOQs, minimum order quantities. They expected more of a financial commitment from us.

Sanjay (23:24): And so, the capital requirements for the company grew substantially. So, at the time that I left, we had raised I think $12 million between the seed and the A-round and then a $5-million note from our existing investors. And then, the company raised an additional probably $80 million or so past that before the company closed.

Sanjay (23:45): And so, the capital needs just got really large. The commitments like the amount of cash tied up in the supply chain became quite big. And then, there are a few cases where the company missed forecast or misforecasted and as a result, there was a lot of cash tied up in inventory. And that created issues around less with the investors but more with folks that you would work with to get debt and venture deadlines to help fund inventory. They're usually happy when everything's going according to plan. As soon as you your cash turns out to be different than you expected, then that can become a different relationship.

Sanjay (24:20): So, what I think happened is that over the last year of the company, there were issues with not having enough cash. And there being this kind bad feedback loop of then not being able to get product out of the supply chain into the channels and selling it. And then that causing more nervousness or consternation from especially the lender groups that were involved with the company. And then that restricting cash flow further and it was this bad feedback loop.

Sanjay (24:49): So, I think at the end, what happened was the lender foreclosed. I think they had pursued M&A trying to sell the company that didn't work out in the end but it was a last-minute thing. And as soon as it didn't work out, they foreclosed to prevent somebody from spending more cash and sold the assets, like sold the IP and everything.

Skateboards to Scooters: Starting Skip

Zach (25:06): So now, you at that time have started your second company Skip. So, why don't we talk a little bit on Skip and both in terms of what Skip was on its own and also how it was similar to and different from a business model perspective from what you were doing at Boosted?

Sanjay (25:23): Yeah, it was very different. So, I had left Boosted in the summer of '17. My co-founder from Boosted, Matt, was starting to work on scooter sharing. At the time, he was saying, "People are going to ride scooters. This is going to be great." And I was skeptical. I was like, "No one's going to ride these scooters. They're going to ride bikes. They're going to ride skateboards but I don't know about scooter."

Sanjay (25:45): He was totally right. And he sat on the corner on this. And the insight that he had was that cities were going to get really involved in how these work. It wasn't like Uber where there's not use of public space. You could imagine giving someone a ride in the city not really even knowing that that was happening. But if the scooter is sitting on the sidewalk, they're going to know and they're going to want to regulate the public space that that scooter is taking up.

Sanjay (26:07): So, the insight that he had along with our other co-founder Mike, they realized they need to work with the cities. And so, their idea while I think this is like right when Bird was first testing their idea in Santa Monica. Their idea was to get a permit to operate a fleet of scooters in a city. And they had convinced Washington, DC which had a really forward thinking DOT, local DOT, to let them try this.

Sanjay (26:34): And some local DOTs are very restrictive about trying new ideas like bike-share or sidewalk delivery robots, things like that. DC's DOT was very friendly to just trying that idea out. And they already tried the bike-share. And so, they were happy to allow them to operate under a scooter permit that was similar to bike-share.

Zach (26:52): So, it's interesting because ... So, Bird was the first one to do the kind of like "let's drop these everywhere and figure it out later" kind of a thing.

Sanjay (27:01): Yeah, that's right.

Zach (27:01): But before that, you'd had I guess what the sort of early versions of micro-mobility is like all the bike-sharing programs with the docks and things like that. And so, what you're describing is a little bit of an in between where we're sort of taking some of the permitting and the permission structure that exists with a dock system, but also the flexibility that exists with a dockless system and trying to find the middle ground. Is that right?

Sanjay (27:22): That's right. I would say the first wave of innovation there was around docks. And that required not just permission from the city but partnership with the city around who's going to pay for the docks, where you're going to install these things and-

Zach (27:34): Right, construction.

Sanjay (27:35): Yeah, construction costs and power routing and everything. Then I would say probably the first folks to do dockless bike-share successfully in the US, especially electric, was JUMP. And they had been running docked biking systems for a while but had come up with this really amazing electric bike that had an integrated U-lock. So, you could lock it to a rack or you could lock it to a pole or something like that. And they were already working closely with cities.

Sanjay (28:04): So, on the bike side, things were very not chaotic. There was a lot of clear partnership between the cities and the operators. And then JUMP was saying, "Hey, here's a product innovation. We'd like to use this in the city environment." On the scooter side, it was, one, the innovation was you didn't have to pedal at all. This is really just like a hop on this thing and go. And it was small enough that you could ride it on the sidewalk which we really couldn't do with the bikes.

Sanjay (28:31): And so, it let people get comfortable on a sidewalk before they went in the bike lane but it also create a lot of chaos on the sidewalk if you're riding on sidewalk. And number two, it opened it up to a lot of people who maybe felt the bike was too much work. And so, the scooter, you could just stand on this thing and go with a throttle and you didn't really have to adjust the seat or get the pedals to feel right. If you were wearing a dress or if you had heels, it might work a lot better than a bike would. So, it opened up the number of users quite a bit and also opened up the chaos quite a bit.

Sanjay (29:00): And so, that combined with "put these on the street and don't ask for permission." That came from a lot of ex-Uber and Lyft folks who had seen that work in rideshare like ask for permission, or sorry, ask for forgiveness later. Those two things together made that grow much, much faster than the bike-share programs do.

Zach (29:18): Yes, grow much faster but then also eventually, that growth got the brakes put on it.

Sanjay (29:23): Yeah. I mean, the city's eventually stepped in and said, "We will require permits."

Zach (29:27): So, you talked about Matt's insight there which turned out to be 100% correct. And at the end of the day, even if they weren't at the very beginning, the cities were going to be the gatekeepers for who's allowed to actually play this game. Did Matt see that? Was that a thought that came up pre-Bird and pre that raised our strategy? Or was it seeing what was happening there and being like, "I think I know where this is going to go. I feel like I know where this is going to go?"

Sanjay (29:55): Yeah, he wasn't seeing that with Bird. I'm speaking for Matt here. But if I remember correctly, I think he saw that with the China bike-share companies because Mobike and Ofo have ... And there are others. There was like Bluegogo. And there are a few other people in China. I think Bluegogo had actually launched in San Francisco.

Sanjay (30:12): And initially, they had launched and the city said, "No." Then they moved to parking spots. So, they would rent a parking spot in a parking lot, just technically private property. And then you'd have to drop the bike back off in another parking spot. So, it was sort of-

Zach (30:24): So, it was sort of Zipcar-ish.

Sanjay (30:26): Yeah, it was kind of Zipcar-ish. But you couldn't just park it randomly on the sidewalk and block the sidewalk. And city had still said, "Nope, we're going to clamp down on this. That's not like an approved use of a parking spot." So, the insight he had was that watching this dynamic happen a couple of times and watching these China bike-share companies try to launch in different US cities. He said, "The US and the Europe will regulate permits for bike-share and scooter-share very tightly."

Sanjay (30:53): So, that was insight number one. And the other insight was that the hardware that people were using for scooter sharing was all this commercial, off-the-shelf scooters that were not designed to live outside. They were not designed for high-duty cycle. They weren't designed for ease of maintenance or swapping batteries out.

Sanjay (31:10): So, the insight was you would need a custom-designed piece of hardware that was designed around security and safety and environmental protection and environmental impact and recyclability. All that would have to be ... Someone has to make that.

Business Model Breakdown: Skip vs. Boosted

Zach (31:25): Right. So, the products that you create when you're doing a shared vehicle is very different because of all those, right? You're maintaining a fleet. And so, you have to think about it differently than a sale of a product like a Boosted Board.

Sanjay (31:39): Yes.

Zach (31:40): The business model and the economics are also fundamentally different because you're...I mean, you're monetizing this thing in a fundamentally different way. The way that that aligns with your cash flow is totally different. So, talk a little bit about the business model differences between Skip and Boosted. And what are pros and cons of those different models?

Sanjay (32:03): Boosted's business model at the beginning was build a piece of hardware and sell it for more than we paid to build it. And over time, we started to look at financing options. We were pretty early ... I think one of the earlier partners with Affirm around financing this product. We found a lot of people that opened up the market to a lot of people who otherwise didn't have $2,000 or $800 to spend on.

Zach (32:30): I think the closest I ever came to pressing the buy button was right when you guys launched the Affirm partnership, and I was like, "Oh, I feel like I can almost justify this."

Sanjay (32:40): Yeah. And the funny thing is we found people who view ... One of the things we found about consumers is that financing a toy feels irresponsible. If you were to get a loan on a jet ski, people would think, "What are you doing?" But a loan on a car sounds totally normal to people. No one would be like you're being financially responsible by taking out a loan on a car tour.

Zach (33:00): Sure. Well, you're looking at how much you spend like how much you're spending on the BART per month then you're comparing that.

Sanjay (33:07): That's not even how they're thinking about it. It's really that there's societal acceptance for financing things that you may be spending too much money on. You might be buying too big of a house or unaffordable car. But if you finance it, that's normal. But financing toys is not normal. There's like what's okay and what's not okay.

Sanjay (33:29): So, financing was something we looked at. And then we looked at ... What was interesting about Boosted was that certain people really wanted the thing to work reliably and we're willing to pay to make sure that parts availability was there or that they could get it fixed and it was going to be working the next day. And so, we started to explore some ideas around like an AppleCare type of program where you could pay to get a loaner board if your board was in the shop or that you could get expedited service if you were commuting on this daily.

Sanjay (33:57): We actually had people offer and say, "I'll pay you extra to get this thing fixed today versus getting it fixed tomorrow." So, we started to explore that. And then, we also looked at resale of refurbished product. And often, an area that people don't think about when they sell consumer products is that something will get returned. And if you offer no return policy, people are really upset. And if you offer a really generous return policy, then you might get a potentially a lot of units back and you have to do something when you can't sell them as new units.

Sanjay (34:26): So, we started to have a refurb program and start to look at where we could use those. We started to look at like fleet sales of selling like 10 or 20. But again, it was very much like a motorcycle. It was not a thing where ... Google wasn't going to buy like 100,000 Boosted Boards to put out around their campus. It was not like that.

Sanjay (34:44): With Skip, it was very different. With Skip, it was there's an asset and we're earning a certain number of dollars each day in ride revenue than we're paying to recharge it, to move it, to process credit card fees, to buy insurance and all the other pay permit fees of the city, all those things. And then, we get to keep the rest. And as long as the amount that we're keeping every day adds up to more than the scooter was worth, now we're making positive cash flow off the scooter.

Zach (35:06): So, what do those numbers look like? What does the revenue per day look like? What does the cost per day look like? How does the math make sense?

Sanjay (35:13): So, today, it's very different than it was at the beginning. At the beginning, there was a novelty factor. At the beginning, people were even ... I remember people were running around trying to find them like it was almost like Pokemon Go trying to find a scooter and then ride it. People didn't really care much about the cost of the ride, like 10-20 bucks to take a ride, who cares? You're going on a date or you're having fun with your friends. People were happy to spend that. They weren't thinking about it as a utility product.

Sanjay (35:39): Today, the numbers are roughly around $10 to $15 per day, closer to $10 in terms of revenue per scooter per day. And that's averaging for summer season. And in winter seasonality, you might get a lot of rides July 4th weekend in DC. But you can't extrapolate that to the rest of the year. You might have a lot of ... We saw a lot of impact if it rained. We'd see huge revenue impact if it was raining. So, weather forecasts were one of the inputs into our model for what our targets were for any given week.

Sanjay (36:10): So, right now, you're seeing like $10, maybe $12, maybe $15 per day in revenue. And then, we were seeing in the best case about 30 cents on the dollar in terms of positive cash flow once you pay for all those other costs, not including the cost of replacing the scooters. And so, steady state, we were expecting maybe around like 15 cents on the dollar in terms of gross margin on the revenue we're making.

Sanjay (36:36): And then, the other limiting knob that you could turn but only so much was how many scooters you could deploy in a city. So, at the beginning, you could put out 100. And often, cities were capping the permits at small numbers to make sure they can manage the program before expanding it. But at some point, you hit like a certain number of scooters per 1,000 people in the city. Maybe that number is higher for high-tourists type of cities like DC. You might have a lot higher tolerance or much higher number of people daily in DC than people who live there. But there's still a limit. You can't just put out a million scooters in DC and expect them all to get ridden two or three times a day.

Sanjay (37:12): So, overall, the numbers end up being about a 15-point, maybe 20-point gross margin on the business and you've got a cap depending on the city that you're in of how many scooters you can reasonably deploy. And then, that divided up among however many players you have. So, if a city can tolerate 10,000 scooters but you've got three operating permits, then your max is going to be 3,300 scooters in that city. And then let's say you're earning 10 bucks a day then you just do the replacement math on buying new scooters in 18 months or 24 or 36 months.

Zach (37:42): So, you're excluding ... When you talk about those gross margins, you're not including the amortization of the scooter over that what it costs per ride based on the replacement or how many rides it can do before its replacement?

Sanjay (37:52): Well, so I think you can get 30%, maybe more than 30% not including the cost of the scooter. And then once you include it now, you're down into the 15%.

Zach (38:00): Got it. Okay. All right, which is pretty tight.

Sanjay (38:03): Which is very tight.

Zach (38:04): And especially I have to imagine city to city variation given that you've got places that are tourist centers. And so, you might expect a certain rider profile there. You've got places that are rideable all year like San Francisco. They're going to have a certain economic profile. Cities that have worse weather are going to have a different economic profile. Less tourism are going to have different economic profile. But also might have very different numbers of people who are allowed to participate that also probably drives the economics a lot.

Zach (38:34): It's been I think recently announced that they were not going to deploy in cities that didn't have limitations on how many companies could-

Sanjay (38:40): Correct.

Zach (38:41): So, the economics there get driven. Early on when you're building your model for what profitability looks like, how right were you in terms of building a financial model here in theory-

Sanjay (38:56): Mostly not. And it was for a bunch of reasons. One was we didn't know where the limits were in each city. It was not just that they said, "We have to give you permission," but we'll also create a bunch of requirements on where you can and can't operate.

Sanjay (39:10): And so, the most profit maximizing decision might be put all the scooters in a touristy area. But the best option for making sure everyone has access to the transportation network might be to put them in non-touristy areas where you're going to earn less per scooter. And so, that tension that exists is I believe why if you look at the history of private systems like the trolleys and the train systems of the United States, eventually they all became public transit. Those were all private systems back 50 years ago.

Sanjay (39:40): And as a result, the city can dictate, "Okay, we're going to make sure that we're serving all neighborhoods with buses not just the profitable routes…

Zach (39:48): The profit-seeking outcome does not match the social goods-seeking outcome. And so, you're either going to end up with like a profitable bus system that does not serve many, many, many people ... or you're going to have a system that serves everybody and is not going to be profitable. And there's no way to solve that problem. That's just the way ... You're going to end up in one of those two places.

Sanjay (40:14): Well, I think what will happen is either the city subsidizes the cost of running those systems, like eventually you want to run them at a loss. Essentially like in order for them to be a subject of social good, eventually they should be free. And for that to happen, they have to be operated at a loss or subsidized by the city.

Sanjay (40:29): But if you're subsidizing a system, it's not a venture-backed massive hit. And so, you have this tension between those two. So, that's one of the things we didn't realize was the cost of we might win in San Francisco permit, but now we might be losing money operating that city.

Zach (40:46): Yeah, that's interesting. It's like the problem with the postal service, the economics with the postal service. But on a city by city -

Sanjay (40:54): Exactly. You have to pick what you're optimizing for. And so, that was one thing we got wrong. Another thing was where density landed like what was the max. So, I think some cities in Texas were much more, "Hey, put as many scooters as you want." But then, all the companies put as many scooters they wanted, you would end up with the consumers choosing which one is the most common because that's the one that's most likely to be in front of you.

Sanjay (41:21): But that company that has the most scooters out is also doing the worst economically because ... I mean, they have a lot of demand but they also have a bunch of unused scooters. They have too many. And so actually having some constraint probably helped which is why maybe Spin made that decision.

Sanjay (41:34): So, there were a lot of things we got wrong about the model. I think it can work potentially like Lyft, for example, is getting a lot of people using a combination of the city bikes or Bay Wheels bike-share systems and Lyft. So, if you think about it as customer or rider acquisition and as a substitute for that, then maybe it works to operate that system at a loss. Or you might be able to put it in a smaller subset of markets like very touristy areas like beach towns, but then you're not going to have the scale of operating it in Brooklyn and all the boroughs in New York or all the neighborhoods in San Francisco.

Zach (42:11): Right. Now, that you've done a couple different businesses with a couple of different business models, what's your perspective on how one should monetize these things? You're building businesses that have stuff. You're making stuff and you're selling stuff along with these services you're providing alongside within like ... Is there a right version for Boosted or Skip or whatever? Is there a right version in general? What's your perspective?

Sanjay (42:39): So, I generally get annoyed by people trying to change the business model to fit what they want and have it not match what the customer wants. And so, that could be forcing you to pay a subscription for your Boosted Board to work correctly.

Zach (42:58): Was there a world where you guys talked it? Was there a future where Boosted had a subscription model?

Sanjay (43:05): I think there could be but it would have to be delivering something of value. It couldn't just be, "Here's a board on a subscription." And I mean if you look at cars, people offer cars on subscriptions but it's generally so much more expensive than just leasing a car.

Zach (43:21): They're crazy expensive.

Sanjay (43:22): Yeah. And there's a reason for that because the cost of running the subscription service and offering everything on top of that and taking on all the liability of the residual value is so high. So, there's a reason that a subscription for something that costs the same as buying it and financing it means like someone is subsidizing that cost, that the true cost is actually much higher.

Sanjay (43:44): So, yeah, we talked about what we could do in that area, but we were always very pragmatic. Whenever we got an investor asking like, "Hey, could you make this a subscription," we always push back. Our discussion internally was always what can we deliver to the customer, and how do we make that something that they would want and where the subscription fits into that.

Sanjay (44:07): The other framework that we learned was there's a list of things that as a consumer, for purely as a consumer product, there's a list of things that you own that if they broke, you would replace them today. And you can make that mental list in your head and that might be your smartphone, it could be your appliances in your house like your front door lock or your refrigerator. If you refrigerator broke, you'd go buy a new refrigerator tomorrow.

Sanjay (44:35): And a lot of products don't fit into that category. So, people buy them or people start to use them, but they're a nice-to-have and not a need-to-have. And then when they do break or when something new comes out, the novelty has worn off and you may not see the same engagement from that user.

Sanjay (44:53): So, once I was having a good conversation with someone, "Would you rather sell 10 products, 10 units of something to people who six of them or eight of them weren't going to use it a month from now or would you rather sell two or four units to people who were?" And on one hand, you're getting all this revenue from those 10 users. And that can help pay for the development costs to get those two or four years of super happy.

Sanjay (45:19): Or you could say, "I don't want to have to build a support infrastructure as a company and the supply chain and everything to support those six or eight casual users because they're not going to be here next year. They're not going to be continued to buy. So, I think both with scooter sharing and with Boosted, our most valuable customers were the ones who needed the product and who really use the product on regular basis. And it's very easy to build for the people who are not using it on a regular basis.

Sanjay (45:50): And you need to be careful because you don't know how people will use it. So, you have to accommodate, you can't force people into that. But whatever the business model like if you just sell something and it's the most basic non-defensible, no subscription whatsoever like the way that a VC investor would think about a hardware product but they're going to use it and they're going to buy new ones like people buy new shoes, people buy new cars. Tesla is not a subscription business. There's aspects of it you could add to it. Or like your phone, you can just buy a phone. The phone is a really good product. And when a new phone comes out, you want a new phone.

Sanjay (46:27): And so, you essentially turn into a subscription customer in that sense. So, I don't think there's necessarily bad models as much as there are products that either have this tacked on or where the product actually isn't fulfilling a real need. And then, that's when things start to fall apart on the business model.

Zach (46:47): What's next?

Sanjay (46:49): Working on new stuff, exploring a project with a couple of friends from Boosted and Skip but not very much in the early stages of figuring out if it will work.

Zach (46:57): Cool.

Sanjay (46:58): Yeah. But it's still hardware-related.

Zach (47:01): Not getting out of the space.

Sanjay (47:03): I thought about it. I mean, I frequently think about it and I think the way that most people who build hardware think about it. We see our friends building software and go, "Oh, this seems easier," or, "At least less stressful. Less gray hair."

Sanjay (47:16): But it's also really fun to build stuff, build physical stuff. So, even just trying to prototype something and watching that come together and then getting to use it is very rewarding.

Thoughts, Advice, and Lessons Learned

Zach (47:30): So, what's your advice to somebody who's at the beginning of the road that you've walked down, either a founder or an aspiring founder or an innovator who's thinking about this stuff at a bigger company and sort of thinking about trying to create the kinds of products that you've created over the last few years? What's your advice for them?

Sanjay (47:50): With physical products, there's a lot of ways for it to go sideways that you won't know at the time. You can think you're going to sell a lot of ... You're going to want to sell a lot of something or you're going to want to make a lot of something, but you might make commitments now that then become real liabilities for you later. And those generally are less dangerous in software. You're not having to pre-pay or pre-commit to your AWS bill for the next three years and the way that you might have to buy products or work with a supply chain partner.

Sanjay (48:29): So, my advice for people early in this is to find ways to keep your footprint small. I think there's ... A lot of people try to build, for example, consumer hardware products like Apple. It's like you can't do that. To do that, you need to have a lot more infrastructure in place. And if so, if you spent a lot of time on packaging, you're spending time on the wrong thing. Whereas Boosted, we found that people who we really needed to appeal to and design for didn't care about the packaging, they care about the product.

Sanjay (49:04): So, my advice would be number one, yeah, don't try to play big company, really focus on making a few people happy which is something that we learned with YC. And the second is really focus on usage, really focus on how people ... Because I think that the siren song of physical products is that everybody was like, "Oh, a new shiny thing."

Sanjay (49:30): And if you think about downloading something on your phone, the cost to let someone try that and get that experience is basically zero. But the cost to you at the early days of building something physical is a lot of work to deliver something to somebody. And if you focus on that moment, instead of are they still using it a week later, are they still using a month later, are they telling their friends about it, the concept of retention and churn and all the things that have become standard practice in software have not really come to the hardware world yet and need to.

Sanjay (50:06): And if you look at really strong hardware companies, they have very strong retention data and very strong retention metrics. So, think about Peloton or think about Apple, those are products that they very religiously measure how people are using them after they buy. And it's very easy when you're building physical stuff to focus on the moment of purchase. That's when you get all your revenue. That's when people are the most excited. That's when the thing is the shiniest. And if you just forget about it after that, you're forgetting about a lot of other things.

Zach (50:37): I think the fact that understanding what customers do with your product after you ship it to them for a physical product is by default so much harder than it is to do that for a website.

Sanjay (50:48): It is.

Zach (50:49): Which once these things become connected, they become more intelligent. Then all of a sudden, you can start to actually learn about how your product is used.

Sanjay (50:56): Yes.

Zach (50:57): But if you're used to not knowing that, then you don't ask the questions that any web service is going to ask about what product engagement looks like. Those things are knowable even for offline products, you just have to talk to your customers and do research and it's hard but you can still figure it out. And it matters even if you can't measure it.

Sanjay (51:23): Yeah. You almost have to be unreasonable about it. If you look at stories from the software startup world, so-and-so went to someone's house and installed their software on their computer or they would go and watch over your shoulder as you use their website. There's like stories like that. And those, you have to do that with the hardware products too.

Sanjay (51:44): And to your point, it's even harder sometimes to watch them use it. But I think it's really important to do that because ... And if you're not getting it right to be intellectually honest with yourself that you haven't figured it out yet. Just because they bought it doesn't mean it was successful. That would be like saying just because they downloaded my app, this is a successful app. And that's no longer the metric for success.

Sanjay (52:05): So, the other piece of advice I'd give to people looking at this is around budgeting. It's very expensive and it's very easy to want to do everything with the physical things you make. One of the wise things we did out of constraints at Boosted was we actually only made a couple of custom parts. Not a couple, but most of the Boosted Board Gen 1 was not made by Boosted. The deck wasn't ours. The wheels weren't ours. The axles weren't ours. We made modifications to make sure that-

Zach (52:40): I was going to say it's surprising because that product feels like a product that's fully-custom.

Sanjay (52:44): It does. And that's some of the magic is to make it feel custom out of pieces that aren't. But for example, the cost to us to go to a custom deck from a deck that we were using from someone else that we already knew, we had to figure out how to reliability test it. We had to get samples of it made. We had to worry about supply chain quality control instead of just saying, "Hey, we want to order 100 decks at a time from somebody already making 1,000 decks of that type."

Sanjay (53:08): And so, constraining the problems you're trying to solve. And I think especially for people who are entrepreneurial and want to create new things, it's easy to "your eyes are bigger than your stomach" kind of happening. But being very disciplined about picking the things to solve at first. And knowing that ... For example, I remember the first time somebody explained to me how to actually calculate cost of goods in a hardware company from a financial perspective included the cost of the warranty that you were going to provide for the product.

Zach (53:44): Even if you don't know actually what the odds are that somebody's going to call that warranty or you have to guess.

Sanjay (53:48): Yes. But you have to think about how many people will buy it and return it? How many people will have it break and need a replacement? Mature companies have warranty accruals. But we didn't know what warranty accrual meant when we started doing this. We just maybe like, "Oh, we just had a bunch of costs come out."

Sanjay (54:03): So, I think especially for Kickstarter projects, it's easy to think about my job is over once I ship it. And actually, your job includes are people using it and do I have that feedback loop? And also, am I taking care of people? And if I say I'm going to give you a return policy, am I now fighting with you over whether I'll fix it?

Sanjay (54:20): And there's definitely companies we've all had experiences with where you're trying to cancel a subscription or you're trying to get something fixed and you're arguing with them over whether they'll honor what you thought was the promise that they made. So, budgeting for that is important.

Zach (54:35): Yeah, cool. Well, thanks for joining us and thanks for sharing your wisdom.

Sanjay (54:39): Yeah, thanks for having me.

On Thursday, May 19 at 9am PT, we’ll be doing a live interview for Atoms and Bits that will be posted in the newsletter a few weeks down the road. The interview is part of Spectra, Particle’s annual IoT conference, and will feature two special guests:

• Zach Shelby, CEO of Edge Impulse (previously: Sensinode, ARM, Micro:bit)

• François Baldassari, CEO of Memfault (previously: Oculus, Pebble)

If you’re too impatient to wait for the interview to be published here, or want to join to participate in live Q&A, register now for Spectra; it’s a free virtual event and we’d be delighted to have you.

Coming up soon in Atoms and Bits: an essay on fundraising for hardware/software businesses and our first video interview with Sanjay Dastoor, founder of Boosted and Skip. Stay tuned!

• You manufacture/assemble your product for $X per unit (including cost of components, labor, overhead, etc.). X is your “Cost of Goods Sold”, or COGS.

• You sell that product to a distributor/retailer/dealer for $Y per unit. $Y-$X is your product margin, which you spend on sales, marketing, and overhead; whatever’s left after you spend that money goes in your pocket as profit.

• The distributor/retailer/dealer will sell the product to an end consumer for $Z per unit. $Z-$Y is the distributor’s product margin, which they similarly spend on sales, marketing, and overhead, pocketing whatever’s left over as profit.

If the price that a customer is willing to pay ($Z) is significantly higher than the cost of manufacturing/assembling the product ($X), such that both you and the distributor can run a profitable business off of that margin, you’re good. Expected margins can vary, but a good rule of thumb is that you want the customer selling price to be 3x or more of your COGS. Z>3X is a decent business, and anything less than that is going to be on the struggle bus.

This business model is being disrupted in a thousand different ways in a thousand different industries. Much like web 2.0 tech (AJAX and friends) that made it possible to deliver true software applications via a web browser opened the door to “Software-as-a-Service” (SaaS), IoT and the introduction of web and mobile software into hardware products is opening the door to “Hardware-as-a-Service” (HaaS) businesses, where companies deliver an ongoing service through a hardware product, and charge a subscription rather than selling you the hardware.

But before we get into a discussion of Hardware-as-a-Service, we have to lay the groundwork by talking about all of the various ways that companies monetize hardware products, and what happens to these business models when you start to inject software into the mix.

For the purposes of this discussion, we’re going to use automotive examples — the businesses that operate in and around automotive sales.

Business model #1: Hardware purchase

As described above, the simplest and most common model for purchasing a hardware product is that it is sold to the customer as a one-time purchase. This is the default business model for most hardware products that aren’t terribly expensive.

Within the automotive industry, a good illustrative example here is buying new tires. If I want to buy a new set of tires, I’m probably going to go to Costco and purchase a pair outright. An average set of tires will cost me between $500 and $700 based on the vehicle I own; I’ll throw that on my credit card, and I will own the tires outright. My relationship with the manufacturer lasts for mere seconds, and unless the tires turn out damaged or defective, I hope I never have to interact with the manufacturer again.

Business model advantages: Payment from your customer is all up-front, so this model often has the best cash flow return to the business. This model is also simple and has very clear delineation of ownership, which has some legal advantages.

Business model disadvantages: While your product might create value for your customer for years, you only have one opportunity to capture value — the initial purchase. This business model also disconnects manufacturers from customers; the manufacturer’s goal is not to ensure that you derive value from the product they sell you, but simply to convince you to buy the product in the first place. The result is companies that tend to be better at marketing than they are at product development.

What happens when you add software to the product? Bad things. “Smart”, software-enabled products rely on cloud services that have to be maintained and mobile apps that have to be kept up to date. Companies that provide software-enabled products but have no ability to monetize that software through recurring revenue will eventually go out of business. This is one of the reasons that Consumer IoT companies have struggled, other than Amazon/Google with their deep pockets. To quote a recent article in The Verge about the death of Insteon (a smart home brand): “any small company that relies on a cloud server, doesn’t charge a monthly subscription fee, and lacks deep pockets, is potentially at risk”.

Business model #2: Financed hardware purchase

More expensive products are often financed, where a bank or other financial institution provides a loan to cover the cost of a hardware purchase, which the customer pays off over some period of time. The loan is typically collateralized by the product itself; if you don’t pay your monthly payments, the bank will repossess the product.

Most of us are familiar with financing through car sales. If I want to buy a new vehicle, I can finance my purchase and pay it off over a few years. Outside of consumer products, financing is extremely common; asset-heavy businesses like factories are likely to be full of financed equipment that will be paid off over an extended period of time.

Business model advantages: Like a traditional hardware purchase, the manufacturer gets paid up front; however in this case the customer gets to pay for the product slowly, which they would likely prefer.

Business model disadvantages: Financing carries the same disadvantages of a traditional hardware purchase, along with the added risk of default.

What happens when you add software to the product? From the manufacturer’s perspective, financed hardware purchases aren’t much different from a traditional hardware purchase, so there is still a fundamental economic mismatch when you add software and services to a hardware product. That said, the bank or financial institution that is providing the loan might be interested in information about the state of the product in order to reduce risk of default; that may open the door to a software/services business, with a major caveat that the customer might not want Big Brother spying on them.

Business model #3: Razor/razorblade

Products with a consumable element may be monetized primarily through the consumable and not through the product itself. This is often referred to as a “razor/razorblade” business model, as it was pioneered by Gillette with the invention of the disposable razorblade. Companies with a razor/razorblade model will often sell their product at a loss or even give it away for free in order to sell consumables.

While you probably don’t think of your car as being “consumable”, many of its parts are, ranging from oil and filter replacements to replacement parts that are sold at much higher margins than the car itself. In recent years, dealerships have opted to offer full transparency on the price of your purchase knowing that there will be little to no profit on the sale of the car itself — but they’ll then focus on profiting off of the maintenance and servicing of that vehicle.

Business model advantages: The razor/razorblade model can be extremely lucrative, given that consumables tend to be very profitable, and in many cases companies can give themselves monopoly power over the consumable business by patenting the consumable. Furthermore, the razor/razorblade model aligns the company’s interests with the customer’s interests; you only make money as long as they actually use your product, so you’re likely to work hard to make sure your product gets used.

Business model disadvantages: Razor/razorblade business models can be ticking time bombs – lucrative up until someone figures out how to steal your consumables business away from you. This is particularly painful if you have been selling your product at a loss with the hopes of selling consumables.

What happens when you add software to the product? Any software/services investments you make in your products are likely to be profitable if they increase the rate of consumption of your consumable and/or increase the odds that the customer will buy consumables from you and not from somebody else. “Automated replenishment” is a common software service for razor/razorblade products and is a common reason to start adding software to ensure that consumables are in-stock and identify opportunities to increase consumption.

Business model #4: Rentals

Rentals are common for products that are used only temporarily and where it is reasonable for a company to take back the product after it’s no longer being used, clean and/or refurbish it, and rent it to another customer thereafter.

Car rentals are extremely common while you are on vacation or while your car is in the shop, where you typically rent by the day or week. Another increasingly common variation on the rental model is “pay-per-use”, where rental fees are based on the utility you derive from the product rather than the amount of time that you use it. This model is popular in micromobility (shared e-bikes and e-scooters), where rental periods are often per-minute or per-mile rather than per-day or per-month. Outside of consumer-land, the place where you are most likely to find rented equipment is a construction site, because construction sites are inherently temporary.

Business model advantages: Strong alignment of incentives with the customer, who is only paying for the product while they’re getting value from it. Generates recurring revenue.

Business model disadvantages: While rentals are recurring revenue businesses, rental periods are typically short and therefore churn is high. Rental businesses have high operational costs to clean/maintain/rental equipment. Typically manufacturers don’t do rentals themselves, as the business of making a widget and renting a widget are very different, and rental customers may want or need variety (rental car companies carry many car manufacturers’ products; construction equipment rental companies carry a wide variety of types of equipment).

What happens when you add software to the product? Rental businesses are a great opportunity to add software to a product, if only to reduce operational costs by monitoring the rented equipment, ensuring that it is well cared for, and making sure you can get it back at the end of the rental period. Anything that reduces maintenance cost and/or shrinkage (loss or theft of equipment) is likely to pay dividends back to the rental company and improve their economics.

Business model #5: Hybrid (Purchase + Subscription)

Manufacturers are increasingly selling their products through a hybrid business model where they sell the hardware through the traditional business models described above and then provide additional value-added services that a customer can subscribe to. An automotive example would be purchasing a car through normal means (purchase or financing) while signing up for an additional maintenance contract where you’re going to pay $X/mo that will cover any necessary repairs or maintenance, oil changes, car washes, tire replacements, even car washes.

This model is traditionally used for products that have a heavy maintenance burden associated with their use. Elevators, for instance, may be sold to a building under a hybrid model where the elevator is purchased and owned outright by the building developer/owner, with a long-term maintenance contract where the building owner pays the elevator manufacturer an annual fee to service and maintain the elevator for the lifetime of the building.

Business model advantages: This can be a “best of both worlds” business model, where the hardware sale covers the cost of manufacturing and delivery of the product while the subscription gives the manufacturer an opportunity to create value for (and capture value from) the customer on an ongoing basis.

Business model disadvantages: Customers might be pissed off about needing to pay for a product “twice”; you can only charge a subscription if you are actually creating enough ongoing value to earn it. Companies who add a subscription after the fact to a product that was once sold outright are likely to anger customers, especially if they haven’t earned their keep.

What happens when you add software to the product? Hybrid business models fit well with software-enabled hardware products, as the customer will perceive that they are purchasing the hardware value in the initial sale and the software value through the ongoing subscription.

Business model #6: Hardware-as-a-Service

The most SaaS-like business model for a physical product is a “Hardware-as-a-Service” business model where a customer will pay a subscription that covers both use of the product as well as associated services. There are a lot of variations of this business model ranging from jet engines sold by minute of uptime to long-haul trucking ELDs sold via a monthly fee. These variations are deserving of their own essay, so stay tuned for a deeper dive here in future issues of Atoms & Bits.

In the automotive industry, this is something that car manufacturers have been toying with, although these initiatives are still very experimental and my impression is that subscribing to a car isn’t particularly popular.

Business model advantages: Maximization of Annual Recurring Revenue (ARR). If you want an example of why this is a good business model, check out Samsara’s S-1 and resulting successful IPO. From their S-1: “We focus on maximizing the lifetime value of our customer relationships, and we continue to make significant investments in order to grow our customer base. Due to our subscription model, we recognize subscription revenue ratably over the term of the subscription period.” This focus on long term relationships - and the resulting long term view on revenue - allows Samsara to focus on continuing creating value for their customers. This aligns incentives and orients them toward products that create customer loyalty and encourage lock-in.

Although Samsara’s stock price has dropped since their IPO in December 2021, that drop is strongly correlated with other tech companies’ stock prices dropping, and they’re still worth close to $7Bn, so I think we can call that a win.

Business model disadvantages: Not charging for a product up-front creates a cash flow mismatch, where the manufacturer has paid for manufacturing and assembly but won’t get paid back on that cost for some period of time. In the current interest rate environment, you can probably find a bank or financial institution to cover this gap, but that’s only feasible when interest rates are low and cash is cheap. Also, as described in more detail below, if you’re going to charge a subscription for your product, you have to earn it.

What happens when you add software to the product? I would argue that this business model is impossible unless the product you provide to your customers is software-enabled; software and “as-a-Service” business models go hand-in-hand.

What is the right business model for you and your product?

With all of that groundwork laid, what is a product company to do? If I make and sell physical products, how should I monetize those widgets? If I’m a hardware/software company, how does that change things?

There is no “right” answer, as different product categories and industries have different customer expectations and constraints. However, there are some basic ground rules that I think everyone can follow.

Those ground rules are:

• Recurring revenue is more valuable than one-time revenue. The ever-increasing valuations of SaaS businesses over the last twenty years prove that both public markets and private investors value recurring revenue at a premium, because recurring revenue businesses tend to grow faster and become more profitable than businesses driven by one-time purchases.

• But — you have to earn recurring revenue, especially subscriptions. Sixt+ will let me subscribe to a car; subscription fees start around $750/mo. That’s a lot of money for a Toyota Camry, which I could probably lease for half of that. Sixt has not earned the right to charge these subscription fees, as they are not providing enough value to a customer on an ongoing basis to justify the premium. As a result I expect the Sixt+ program to either change dramatically or disappear.

• Hybrid business models are often a good compromise and a good transition to an “as-a-Service” business model. Going from a traditional hardware sale to a full subscription business model is a heavy lift; hybrid business models can create a lot of flexibility for businesses to create additional value over time through software and ongoing services, and as that part of their offering strengthens, shift from purchase-heavy revenue to subscription-heavy revenue.

• Traditional hardware purchases do not support software investments. Companies who are investing money to create ongoing value for customers through software but aren’t charging customers for that value either need infinite cash to support those investments (Amazon/Google) or those products will get shut down (Insteon, Revolv, etc.)

Phrases like these don’t get a lot of critical thinking. If you run a business that manufactures a physical product, “hardware is hard” is the kind of thing you’d say in a meeting while lamenting the unavailability of silicon. Or when complaining about a delay. Everyone around the room will solemnly nod and agree.

Hardware companies face unique challenges that pure software companies don’t. Inventory, supply chain, shipping, logistics. Slower design cycles. Defects and returns. There are many challenges associated with making real things that software companies don’t have to deal with.

But there are lot of things about software that are hard too. Pricing is hard. Reducing churn is hard. Increasing sales velocity is hard. Recruiting is hard. Scaling is hard. Is hardware uniquely hard?

I would argue that the biggest difference is mindset. The challenges a software company might face are perceived as solvable problems that separate strong SaaS companies from weak SaaS companies. Some software companies have high churn; others don’t. If your churn is low, you’re good. If your churn is high, fix it or die.

We’ve developed this kind of mindset because the internet is full of great stuff like this:

If you are building a SaaS business, there is an incredible wealth of shared knowledge available to send you down the right path. Investors know what to look for in a SaaS business; founders know what metrics to monitor and improve; executives know what problems to fix and what opportunities to exploit. This knowledge is quickly and easily passed around through Medium posts and tweetstorms and books and conferences and 1-on-1 mentorship and seasoned executives who bring experience from company to company.

There is no such shared knowledge for hardware businesses.

This is what makes hardware hard. It’s not capital intensity or gross margins or operational challenges. Those are all solvable problems. It’s the lack of shared knowledge among founders, investors, and executives regarding these challenges and how to overcome them.

So, let’s fix that.

My name is Zach Supalla, and I am the founder and CEO of Particle, an IoT Platform-as-a-Service (PaaS). In addition to our cloud service, my company makes hardware — modules and gateways that connect to our platform. And our customers include thousands of OEMs, service providers, and start-ups who design and manufacture some kind of internet-connected widget. I’ve spent the last decade learning the ins and outs of how companies make stuff, and how they build profitable, differentiated, and scalable businesses while doing so.

What I haven’t done — what none of us have done — is write down what I’ve learned. The hard-earned lessons, the mistakes made along the way, the experiments that succeeded and failed.

This newsletter — named for an oft-used phrase describing businesses that combine hardware (atoms) and software (bits) — is intended to be a resource for people leading or investing in what I will call “integrated hardware and software businesses” — “HW/SW businesses” for short. I’ll  answer questions such as:

• What separates a “good” HW/SW business from a “bad” HW/SW business?

• How can HW/SW businesses manage and reduce the costs and risks associated with their hardware products?

• How do you price a HW/SW product?

• How does a pure hardware business become a HW/SW business? How does a pure software business become a HW/SW business? Should they?

• Which SaaS metrics can be applied to HW/SW businesses? Which can’t?

Some of these questions I can answer on my own. For others, I’ll be bringing in experts — founders, executives, and investors who play in and around this space. Atoms & Bits will alternate between essays like this one and video interviews with other people who know things that I don’t know.

If you’re interested in following along and seeing what I have to say, subscribe; we’ll be publishing twice a month. And if there are particular questions you’d like me to answer or people I should interview, shoot me a DM — I’m @zs on Twitter.

Stay tuned for the first real issue, coming up in two weeks.

Subscribe now