The Real Goal of a Startup: Get to "Default Alive"
Watch on YouTube, or listen on Spotify and Apple Podcasts.
Mash sat down at Wayne Fenton’s home on the Santa Cruz coast to talk about the stretch of a startup’s life that nobody romanticizes: the desert before product-market fit. Wayne spent decades inside companies big and small — Oracle, Cisco, eBay, Qualcomm, Upwork — before becoming a first-time founder at LightLinks, a Santa Cruz startup using light to move data faster than Wi-Fi. Along the way he mentored at Santa Cruz Works and taught entrepreneurship at UCSC. The conversation gets into the rule of four, why a technology looking for a problem can work after all, customer discovery that isn’t a survey, and why every stage of a startup has a cliff at the end.
This conversation has been edited and condensed.
“Join a startup. Don’t launch a startup”
The last big company you were at was Upwork. How did you get there?
I started at a big company, Oracle. Then I tried a startup, and it was fantastic — we had an IPO, and then we got bought by Cisco. So I was back at a big company, decided that was too bureaucratic, and went to eBay as a software architect.
Then I was VP of engineering at a startup. We were kind of mediocre, but we got bought by Qualcomm. So I was a VP at Qualcomm for several years working in IoT, which is foreshadowing of this LightLinks world. From Qualcomm I went to another IoT startup, a medium-sized company, that got bought. Then I went to two failed startups. I literally got fired from two failed startups — although they failed anyway without me.
Then I went to Upwork as an individual contributor, a software architect, because I was sick of being a VP dealing with HR issues. Later I went back to managing as a director, and that was five years or so. During Upwork the pandemic happened, we went remote, and that’s when we moved over here to paradise.
A lot of students come out of university wanting to go straight into the startup world. You built an entire career of experience working in and around startups first. What do you recommend for someone coming out of college?
I usually recommend kids, if they want to learn or live the startup world: join a startup. Don’t launch a startup. It’s like having sex. You can’t learn in school how to build a company. It’s like having a baby, or dying — these things are just not learnable except by doing them.
I always picked startups that had maybe 20 to 50 people. The one really successful one, I was employee number 11. But generally they weren’t super early, and they’d shown some progress toward product-market fit. That’s the key, and it usually triggers an A round. If you’re in seed or pre-seed, you have no evidence yet. So I always targeted Series A startups or around there, because there’s evidence. When you hear that 90% of startups fail — by the time they get to the A round, it’s not 90 anymore. It’s 50.
It’s really hard to get a unicorn, so find one with a medium amount of risk and a medium amount of reward. I came up with a baseball analogy. A single is an acquisition or an acqui-hire, a double is an acquisition for $10 million, say, then a triple, then a home run. I had a very high on-base percentage, but I never had a home run — three doubles, a triple and a single, something like that, and two strikeouts.
That’s critical, because people have an opportunity cost. As a student you can go get a job, or choose a startup — and then which startup? Join a dorm-room buddy as a co-founder and you’re in that 90% failure rate. In all those startups, were you ever a founder?
No. This is my first one, and it’s probably my last one. Founding is impossibly hard. If you knew how hard it is, you would never do it. You learn by doing, and then you can repeat it and you get the lessons.
Investors will also more readily invest in a repeat founder. They hate first-time founders for this reason, and this is why LightLinks struggled: my co-founder, Firouz Vafadari, is a first-time founder. He did a little company in Iran, but in America this is his first startup. Now we’ve got a big customer, we’ve got evidence, we’ve got me, we’ve got a COO.
“That’s not a problem to be solved”
You moved to Santa Cruz during COVID and landed in the middle of a new ecosystem, with its own Santa Cruz flavor. How did you get involved?
Two things. I’d been teaching programming over the hill for UCSC Extension — Python, C — so I’d already been vetted and approved as an instructor. And I wanted to do some volunteer work, and startups were perfect. I learned about the UCSC ecosystem and the Santa Cruz Works innovation hub. They work together, and they bring in local tech companies and local businesses.
So I started poking around. I became a mentor at Santa Cruz Works for their little accelerator, and I started volunteering at the entrepreneurship events on campus: hackathons, ideathons, pitch competition workshops. I met a provost on campus who asked me to teach a class, and the class was how to operationalize and scale a startup. That was perfect, because I was already trying to formalize this flow: what is a good idea, what is the problem to be solved, what is the ideal customer profile. You read about them in books — The Lean Startup, and one I’ll talk about later called Pattern Breakers — and they’re all nuggets of this. But I wanted to get it clear, and the class was a good excuse.
On the side I’ve been mentoring startups through UC Santa Cruz’s flow: an ideathon in the fall, then hackathons, a pitch competition workshop, and a pitch competition in May. A lot of those people then flow into Santa Cruz Works’ accelerator.
And that’s where you met your co-founder?
I think we met in the pitch competition workshop. He liked my advice and asked me to advise them going up to the pitch competition. The elevator pitch is using optical light to get higher-speed data transmission — more bandwidth than Wi-Fi. But he comes to me and his idea is, “We’re going to replace Wi-Fi.” And I’m like, that’s not a problem to be solved.
If you want to replace an existing technology, have you heard of the rule of four? It’s simple. If you’re going to replace an existing technology, you can’t be 50% better or two times better. You have to be at least four times better. I don’t know why it’s four or where they came up with it, but it’s easy to remember, so I use it. And I said, you’re not convincing me what problem this is four times better than Wi-Fi for.
He said, “Well, we’ll figure it out.” He went through Santa Cruz Works, and they helped with customer discovery and things like that. Then they applied to Berkeley SkyDeck, and shockingly they got in. We got $200K of seed money. I think it’s mostly to help the kids live while they go through the six-month process.
What’s the difference between a local program and SkyDeck?
Santa Cruz Works is smaller, with local mentors and some curriculum, and it’s helpful. When you get to Y Combinator or SkyDeck, first, you get money — for legal bills, patents, paying people. Ours is a hardware product, so we had to buy things to make prototypes and MVPs.
The other thing about SkyDeck is there’s a massive network: 800 mentors. You go to Chon Tang, the director, and say, “I need an expert in the wireless defense area.” Here’s four names, and they will all answer the phone or the email. Or, “I need two investors in the deep tech space.” Here you go. You probably won’t get the money, but you get those meetings, because they know SkyDeck and they’re sympathetic. That’s the difference between a small scale and a huge scale. They have 2,000 applicants for 20 slots each cohort. It’s very hard.
We got in, we went to demo day, and we started trying to fundraise. But we were too early. We had no evidence.
Even SkyDeck let you in too early.
Because of the technology — it’s clearly got to be useful somewhere. People always say, “That’s just technology looking for a problem. Never do that.” I’m now convinced it’s okay, as long as you find the right problem. And it’s hard. From May until August, Firouz was just probing, using those connections: industries, use cases, who is having problems with wireless communication.
In August he got a connection to a massive company, in their innovation program, and they said, “Wait, we do have a problem.” They’re automating warehouses. Robots are everywhere, and they’re redoing all of that now with AI. The issue is you have to coordinate the vision system with the arm at high speed. On a car line, those robots are connected by cables and they’re fixed. But this new generation of robots moves around, so they have to be wireless.
The problem with Wi-Fi is tens of milliseconds of latency. That doesn’t sound like a lot, but it’s very variable, because it’s a high-congestion environment. It goes 30, 40, 50, and spikes to 100 milliseconds. When we saw this in action, you could literally see the arm stagger as it’s trying to get over to the product.
They’d tried 5G and all these different RF solutions. They told my co-founder, “Get on a plane. You need to be here by Monday morning.” He packed up his Pelican case and went. Our solution looks like a normal network device, so they turned it on, it connected, they started the test program, and it was completely different. It’s like 10 times faster than Wi-Fi.
So it’s got the 4x — the order of magnitude.
Exactly. Companies like that would never work with a startup like us. This was a weird, janky prototype. And they said, “Well, we’ll let you try.” So we’re in a pilot program right now.
“A cliff at the end”
Deep tech coming out of universities often falls into “that’s cool that you invented something — but who’s going to use it?” Universities have tech transfer programs that help researchers patent their technologies, but there’s no nice handoff to commercialization. I think you need both. Innovation without a pathway to commercialization is just really cool stuff you get awards for. And teaching kids entrepreneurship without anything worth selling is missing the other side.
Totally. If you’re building an app, fine — you put it out there, maybe you market it or get influencers. But we’re building an enterprise-class networking product. Who do you even sell to? Does my co-founder know about enterprise sales processes? No.
Another one I see a lot is edtech: “Let’s improve education with AI.” Who are you selling to? The school district has IT systems with security approvals nine miles long. Commercialization is very different depending on the industry and the target customer. How do you standardize that? Maybe it has to be advisors and mentors, because that’s the only way you could know.
SkyDeck evolved over time. A lot of universities start out trying to own the venture, the ecosystem and the education all at once, and then the advancement people seeking donors fight with the venture people trying to get those same people to invest. SkyDeck separated that: a venture fund outside the university that returns 50% of its carry. And LightLinks came out of UCSC’s programs and competitions and into what was available for free in the community — no money, but it kept you alive long enough.
Yep — more feedback, talking to different people in that group. And this is what breaks my heart. Each of these phases that are separated has a cliff at the end. They either fall off the cliff and disappear forever — from the UCSC competitions, they just fall off — or they go get regular jobs. They give up and say, “Well, that’s too hard.”
So what’s wrong with our system today? Why don’t we put out more startups?
I had this question too. We have all these frameworks now — jobs to be done, customer segmentation, a business model matrix — and they’re helpful, but they still don’t change the failure rate that much. I was in a webinar with Jason Calacanis. He’s kind of a gadfly of the entrepreneurship world, he runs an incubator and an accelerator, and then he invests in them himself with his own syndicate — cradle to grave. In the Q&A I asked him why the failure rate is still high.
He said it’s just too hard to standardize. You can get the steps, and you can help kids avoid the failure modes that are obvious, but building a product that people will pay for and scaling it is still just hard. Even with all the evidence milestones — what do you need to get your Series A, what do you need to show — it’s just freaking hard.
There’s an element of magic. If you smash enough people together, something happens. Then there’s just luck. And like you were saying, on the path it’s just: don’t die.
Right. Just don’t die. You have to get to default alive. That’s the journey. You’re wandering out in the desert, and you’ve got to find product-market fit, because you need your customers and your market to pull. You can’t keep trying to jam a product into people with pilots and evaluations and little betas. You’re on a journey to product-market fit so you can be default alive, so you can then hopefully get into a growth mode — a hypergrowth mode. That’s the idea.
“Just don’t die. You have to get to default alive.”
— Wayne Fenton · Inspiring Founders Podcast, Ep. 05
“Customer discovery is not a random walk”
Do you have any tricks for getting there?
Three things. The first I said earlier, but I want to make it clear: I now think it’s okay to identify a technology and go looking for a problem — as long as you find a good, hard problem you can solve, and it’s a problem you want to dedicate 10 years of your life to. In most startups, the founders just quit.
So it’s not a technology looking for any problem. It’s within a problem space. It still replaces Wi-Fi — it just needed to find the people who were having the problem with Wi-Fi.
Exactly. The specific use case within that problem space. That was our journey, and my co-founder spent six months probing.
Here’s what I’m learning about customer discovery. When you go into an accelerator, here’s the process: you do interviews, you have standardized questions that let you evaluate whether they’d pay for this and how much. The problem, particularly in deep tech, is that’s not how you find the problem. We went into defense, we went into sports stadiums, and we only got into those discussions because of the network. He was able to go from use case to use case, get answers to emails and have meetings, until he found that specific one. That was real customer discovery. It’s like Moses in the desert, just wandering. It was organized — here are two different people in defense it might work for, but it’s really hard to sell there.
The book I wanted to mention is Pattern Breakers by Mike Maples. A lot of successful companies take an inflection point and apply it to a problem they can now solve differently. You look into the future and ask, what can this inflection point that’s just happening solve?
His example is the iPhone 4, which came out with GPS and a GPS API for apps to use. That’s the only thing that enabled Uber to exist. My example is my first Uber ride, leaving a party on a Saturday night. I figured it was probably going to be 20 minutes. I called it, looked down, and it said, “Your car is arriving.” That was an inflection point that enabled a whole thing.
I’ve worked with thousands of students, and almost every one of them asks, “Mash, is there any other way I can do this? Can I send a survey? Can AI do it for me?” And I’m like, get out of here. Go talk to somebody.
You can’t learn the pain. How painful is this problem to them? You can’t even really ask that, except by sitting there watching them have the problem, like we did. We sat there and watched this thing and we’re like, whoa, that’s not working.
So customer discovery is not a random walk, and that’s how you get a real ideal customer profile. Now that we’ve done one of these, it has to be a wireless problem, and speed has to be critical. It helps you really narrow your target.
“You have to get evidence early”
What’s the second thing?
What I said earlier: I was surprised that every step is perilous, every step is required, and it’s serial.
What does that mean for a new founder who has an idea and wants to get to where you are? Should they be selective about the programs they spend time in?
Not everybody needs an accelerator. For deep tech, or something like an enterprise sale, I would argue yes, particularly if you’re a first-time founder. For us, an early acknowledgement in a pitch competition helped us get into a free accelerator, which helped us get into a paid accelerator. So I can enumerate all those steps. But I can’t automate it.
How do you know you’re ready to go to the next step — from a hackathon to an incubator, say?
One example: there was a kid who wanted to do handyman-as-a-service with students in Santa Cruz. He won a hackathon, and he had some people doing it — I used him, and he was great. Then he graduated. I was asking him, “What are you learning about the labor pool? This is a marketplace. You need two-sided participants and enough critical mass, like in the Uber case.” He kind of withered, because he couldn’t scale that. I don’t think an accelerator would have helped. Other than mentorship with people who have done it, I don’t know how you know you’re ready.
Maybe it’s about learning what kind of company you are, as quickly as possible. He could have built a fine lifestyle service company. But if everyone’s telling you, “We could take this global,” and you go down that path, it’s a fool’s errand.
That point is the good one: you have to get evidence early. If you don’t get evidence, all hope is lost. Maybe that’s the thing about the gates — you need the evidence at that gate or you’re wasting your time. Each of those stages has a set of evidence you can collect or present as you go. This whole Silicon Valley thing is an evidence-based, milestone-based sequence, and you can’t skip any of it.
Which might be our responsibility as ecosystem builders: not to say “you can’t pass this gate,” but to help founders understand that what they’re actually doing is de-risking the company at every stage, and the milestones come with that. There are outliers who win a hackathon and hit $50 million ARR three months later, and everyone loves them. But for most of us there are milestones to meet, and being willing to do it transforms who we are through the process.
That, and it’s hard because people pivot. “Okay, I acknowledge this isn’t going to work, but I’ve got this other idea.” Like Slack: “I was doing a gaming product, it was flailing, but we made this communication tool while we were doing it. Let’s see if there’s interest.” So when something fails at a gate, it’s not necessarily throw in the towel and go home. But you have to be open: this isn’t working, the evidence isn’t there.
“Awesome for research”
I have to ask, because it’s the biggest topic in the world right now. How is AI helping LightLinks — not necessarily in the product, but operationally?
First, AI in the world is the market we’re targeting, so there’s that. As an internal tool, it’s awesome for research — really fast experimentation on how this could be used in a given use case. It has massive amounts of perspective: oh, you could do that underwater, in space. So it’s great for pressure-testing ideas.
It’s critical for IP generation: generate a draft patent out of this idea. Generate a mutual NDA. Help us build a financial model. It’s getting a little better at presentations, although that’s still pretty bad — but it’ll generate an outline and then I can go update it.
Another example: do we need to hire a firmware developer, or can we ask AI to build these firmware elements? I’m a programmer and I did a little bit with IoT, but that might be an area where, if you don’t have the expertise, you can augment.
More conversations with founders on the Inspiring Founders Podcast.