Lessons learned at Vouch

Lessons learned at Vouch

Two years ago, excited to begin a new journey, I called into an onboarding Zoom from the makeshift desk I set up in my kitchen at the start of the pandemic. I had just left Apple to join Vouch, a startup that sells business insurance for startups, and it felt bizarre to start a new job from my apartment, not knowing when I might meet my new coworkers in person.

When I joined, Vouch was about 70 people, half of them engineers. Today, the company has grown to over 200 talented individuals, and Vouch is well on its way to becoming the leading business insurance provider for innovators.

I joined Vouch with a very specific goal in mind: I saw it as a stepping stone to starting a company. Working there was an opportunity to pick up the right habits at a fast-growing startup. Before Vouch, I was mostly siloed in giant engineering and product orgs, and I wanted a taste of interfacing with nimble go-to-market and scrappy administrative teams. More importantly, I wanted exposure to diverse businesses and their unique challenges.

There was no question that data was going to be key to the company. My main responsibility was to set up the company’s Data org, building data systems to help the broader team find patterns of what makes or breaks companies, and to measure our own performance over time. As expected, the role gave me a holistic (albeit secondhand) view of the startup ecosystem.

Here are some of the key takeaways from my experience at Vouch:

Early decisions are sticky, and they often lead to unintended consequences. The boundaries you set between teams, the tools you choose, the early infra you build, and the experiments you run all snowball and shape your team’s understanding of what is possible as a company grows. This is also true in larger organizations, but it is obvious in startups. For example, I can clearly trace how flowing incomplete but real-time insights into Slack channels later undermined people’s trust in more robust dashboards, or how the concepts we created in our data warehouse caused conflict as systems and teams’ mental models drifted. On the other hand, the early choices we made in our entity resolution system shaped how easily we could incrementally ingest and deduplicate new data. Path dependence goes both ways.

Tech companies accumulate knowledge into software and data products, creating value by layering many processes together. At startups, layering on a blank slate affords many iterations to cheaply walk back mistakes, but at some point you have to work on top of bad decisions and move forward. A lot of ink has been spilled discussing technical debt, but after my experience at Vouch I’m convinced that business processes evolve the same way. In fact, tech debt might always have to be evaluated through the lens of business operations debt, because by the time you clean up your mess on the tech side the business might have moved on, or died. Paying down tech debt is a privilege.

It’s too easy to pick the wrong problem to solve. Probably the hardest thing about working at a startup is that the opportunity cost is much higher than at an established company. In startups, there’s no redundancy. Tough decisions await around every corner: do you build half-baked ideas, risking rework, or do you delay to seek alignment? Do you design for traffic you hope will materialize later, or do you build features that could deliver that traffic in the first place? Do you try to work with the unresponsive but high impact team, or do you continue working with engaged stakeholders whose projects won’t move the needle much? Since there’s less structure, leaders must define how they want to constrain their teams, and those prioritization decisions can be make-or-break.

For example, my team spent a lot of time building tools and pipelines to merge external datasets with our own data. I’m of course biased, but I’d be surprised if there was a better startup traction dataset out there than the one my team cobbled together from various vendors’ and proprietary datasets. In retrospect, this was bad prioritization, though. We spent most of our time deepening coverage, or improving the quality of individual attributes. However, without embedding our more reliable data points in their standard workflows, users just ignored our platform. Unfortunately, we didn’t figure that out until way after we had made the bulk of our investment and our users had gotten used to the worse tools at hand (again, decisions are sticky). It is hard to manage the software development life cycle once you know what needs building, but the real challenge begins even earlier.

Team building is culture building. I helped build three teams at Vouch: Data Engineering, Enablement Engineering, and Analytics. I was also part of the committee that redrew our company values around the time we crossed the 100 person mark. Culture was always top of mind. With bigger teams and broader responsibilities, leaders need alignment while no longer having full control, and culture is the main tool to do that. I had hired people in my previous role, but those people were coming into a fairly set culture. At early stage companies, each hire has an outsized impact because we model our behavior after each other, and early on there aren’t many other people to mimic. The people you choose bring their values and ideals into the company, defining how your team operates.

Teams can be empowered to make decisions and rally behind shared objectives, but sometimes those objectives conflict with each other by design. Selling more premium sounds great to the sales team, but the underwriters are also keeping an eye on the loss ratio, and it’s easy to move one at the expense of the other (not too different from the opposition of engagement and coverage rates in my old life in Search). But people pay attention to what’s celebrated, incentivized, and discussed, which is ultimately all that culture is. And culture applies at a micro level, too – even when I just had a team of six and a few dotted line reports I sometimes felt I had no grasp of many things in flight. Beyond programmatic checks for the basics, the only way I could keep things in line was via trust and culture: setting the example with my actions, and being deliberate about what I rewarded and chose to discuss.


While many of the learnings I shared above can be read as negatives, that’s because challenges are what taught me the most. If I could go back to early 2021 I would make the same choice and join the company again.

Too many people to name made my time at Vouch special. I am particularly proud that the data team I built moved at a pace I had never experienced before, all while treating each other respectfully, including our business stakeholders and the engineering teams we interfaced with. We built a lot together, and I look forward to seeing what else that team builds next. All in all, I’m glad I was part of the Vouch story.

As for what I’m up to next, you won’t be surprised that I’m starting a startup! Leaving the comfort and excitement of a successful startup like Vouch to embark on this new entrepreneurial adventure will require a mindset shift, but it’s one I’ve been expecting for years now.

Stay tuned for more!

Thanks to Hannah Doherty for her feedback on early drafts of this post.

Photo: Window Cleaners, by me. Previously posted on Work Travel: Chicago and Columbus.

Want to see more articles like this? Sign up below: