Ready to build your own Founder-Led Growth engine? Book a Strategy Call
Frontlines.io | Where B2B Founders Talk GTM.
Strategic Communications Advisory For Visionary Founders
Before pitching institutional investors, Sean needed answers to two questions: will customers actually buy home insurance online, and can Kin's algorithms outperform legacy data collection on home traits? Rather than relying on surveys showing 70% of customers prefer buying online, he bought a small existing broker and ran real marketing experiments against it — getting actual conversion data, not stated preference. Simultaneously, he and his co-founder knocked on APIs and trained basic image recognition models against public data sources to test whether machine-generated home data could beat the industry's bar of asking a middleman. Both proved out. The sequencing matters: run cheap real-world experiments against your two biggest unknowns, prove them, then raise. It changes the nature of the fundraising conversation entirely.
The insight Sean drew from studying Capital One: in any business where you're pricing risk, who sees your ad determines what your portfolio looks like — not just your conversion rate. Legacy insurers run untargeted Super Bowl campaigns and appoint local agents, which works for Fritos or Budweiser where everyone in the audience is a potential customer. For insurance, you're spending heavily to reach people who either can't qualify or aren't in market. Kin's model ties every ad bid to a per-address expected value calculation — factoring in risk profile, predicted conversion probability, and lifetime revenue. The bid might be zero, it might be five dollars; it's determined entirely by what that specific address is worth. Marketing and underwriting are not separate functions at Kin. Founders building in any risk-sensitive category — lending, healthcare, insurance — should operate the same way.
Two years in, Kin's carrier partner was acquired and the new leadership deprioritized the relationship. Kin had built its entire operation on that partner's regulatory infrastructure and capital — and suddenly couldn't grow under the old arrangement. Declaring independence required simultaneously securing equity, debt, and regulatory approval from the state, with all three contingent on each other. They ran out of room: one week of runway left when it finally came together. The specific lesson isn't just "avoid single points of failure" — it's that in regulated industries, any partnership that controls your license to operate is a structural vulnerability. If you can't independently obtain regulatory approval and capitalize your own balance sheet within a defined timeframe, you don't control your own company.
Insurance is regulated state-by-state, and regulators have long institutional memories. Sean was explicit: he had no idea how to start a regulated insurance entity, so he hired people who had spent careers navigating exactly that. Those hires came with existing credibility with state regulators, membership in trade organizations, and professional networks that enabled Kin to co-draft legislation with other carriers to close fraud loopholes. That regulatory credibility directly affects speed-to-market — states where you have established relationships move faster. In regulated categories, your legal and regulatory team isn't overhead; it's a competitive advantage that compounds over time.
With 200,000+ customers and nine years of claims data, Kin is now expanding into auto insurance and home equity financing — HELOCs and mortgages. Sean is clear-eyed about the competitive position: Kin probably won't have the same technological moat on these adjacent products that it built in homeowners. But it doesn't need one. The homeowners relationship already surfaces the customer's risk profile, creditworthiness, and home equity position — the exact inputs that underwrite auto and home equity products. The first product built the data asset; the adjacent products monetize it. For founders thinking about product expansion, the question isn't whether you can build a moat in the new category — it's whether your existing data makes you a better underwriter or seller than a cold-start competitor in that category.
There are 90,000 bank branches in the United States. There are 400,000 local insurance agent offices.
That ratio — five insurance storefronts for every bank branch — tells you almost everything you need to know about why homeowners insurance is ripe for disruption. And it’s the gap that Sean Harper, CEO and Co-Founder of Kin, has spent nearly a decade closing.
In a recent episode of Unicorn Builders, Sean walked through the GTM journey that took Kin from a two-man experiment in Chicago to a direct-to-consumer insurance company serving more than 200,000 customers across more than half the U.S. The story is equal parts market thesis, near-death experience, and a lesson borrowed from a credit card company that most people wouldn’t think to apply to insurance.
Sean didn’t set out to build an insurance company. His background was fintech — his previous company was a payments business — and after a small exit, he gave himself a year to map every financial product category and find where technology was genuinely absent.
Home insurance kept surfacing. More than 70% of customers say they’d prefer to buy it online. Almost none do. But what really caught his attention was how the industry collected data on the homes it was insuring: it asked the homeowner, or it asked the agent.
“The legacy insurance companies just didn’t know enough about the homes,” Sean said. “They’re basically pricing them all as if they’re the same when they’re not.”
That’s not just a product problem — it’s a pricing arbitrage. If you could build a model that actually understood home construction quality using satellite imagery, MLS data, tax records, and building permits, you wouldn’t need to beat 100 years of incumbent data science on day one. You just needed to clear a very low bar. And the bar — self-reported data from a middleman — was low enough that basic image recognition algorithms could clear it from the start.
That realization shaped Kin’s entire technical strategy.
Before raising institutional capital, Sean needed real answers to two questions: would customers actually convert online, and could algorithms outperform the industry’s data collection method?
The important context: Sean had prior exits. He had credibility. He could have raised a seed round on the idea alone. He chose not to.
Instead, he and his co-founder bought a small two-person insurance brokerage in Chicago and ran real marketing experiments against it — not to build a product, but to get actual conversion data instead of survey responses. Simultaneously, they spent months training image recognition models on public data sources to test whether machine-generated home data could beat the industry’s bar.
“The fortunate thing for us is both those experiments turned out to be true,” Sean said. “We could get users and we did have more accurate data than the very low bar that the industry has, which is asking some middleman about the house.”
With both assumptions validated, they raised their institutional seed round at month twelve. For second-time founders with the credibility to raise earlier, this sequencing is a deliberate constraint — and it changes the nature of the fundraising conversation from “trust us” to “here’s what we’ve already proven.”
As Kin built its marketing motion, Sean kept returning to one reference point: Capital One.
“The thing that they did really well was they understood that data was the key to winning the risk game and the marketing game,” he said. “The marketing was actually one of the biggest levers for creating good risk selection.”
The mechanism is precise: in a risk business, who sees your ad determines what your portfolio looks like — not just your conversion rate. Legacy insurers run Super Bowl campaigns and rely on local agents to convert them. The fatal flaw isn’t just that the ads are untargeted. It’s that when a customer does convert through an agent, the carrier never gets the data back. They have no visibility into which customers responded, which didn’t, and why. Marketing and underwriting are completely decoupled — which means they can’t improve either one.
“They actually have no way of even measuring the response because the response is happening through this channel they don’t even have access to,” Sean said. “They don’t know which customers converted and which ones didn’t.”
Kin built the opposite model. Every ad bid is calculated at the address level — factoring in risk profile, conversion probability, and lifetime revenue. The bid might be zero. It might be five dollars. It’s set entirely by expected value. Last year, Kin spent $100 million on customer acquisition — all direct response, no brand advertising. The targeting decision and the underwriting decision are the same decision.
Nigel Morris — who founded Capital One and later invested in Kin through QED — described the company as “the Capital One of insurance.” Sean’s response: that was the whole point from the beginning.
The hardest GTM lesson from Kin’s journey has nothing to do with marketing or product. It’s about how you structure dependencies in regulated industries.
In Kin’s early years, the company operated under an arrangement with a legacy carrier — the partner provided regulatory infrastructure and capital; Kin did the distribution work and collected a fee. Then the carrier got acquired. New leadership, new strategy, and suddenly the relationship that had enabled Kin’s growth became a constraint on it.
“We sort of found ourselves in a spot where we weren’t really able to grow under our old relationship,” Sean said.
Declaring independence meant securing equity, debt, and state regulatory approval simultaneously — with each contingent on the others. The trigger wasn’t a strategic mistake Kin made. It was an M&A event they had zero control over. That’s the precise lesson: it’s not enough to trust a partner relationship. The question is what happens to your ability to operate when their ownership changes — and whether you’ve structured a path to independence before you need it.
They got down to one week of runway before it came together. The regulatory approval arrived. The first investor wired. Then the others. Then the debt. They called it Kin Dependence Day. The Declaration of Kin Dependence still hangs on the wall.
The Data Moat That Took Nine Years to Build — and What It Unlocks Now
In the early days, Kin had a thesis about input data and no output data. Legacy carriers had the opposite problem: flawed inputs, but 100 years of claims history.
“These companies are sitting on 100 years worth of output data,” Sean noted. “So they have bad input data, but a ton of output data.”
Nine years later, Kin has validated the thesis on both sides. The machine-generated home data outperforms self-reported data. And the claims history is now deep enough to be statistically significant. That combination took nearly a decade to build and is increasingly difficult for a new entrant to replicate.
Now Kin is expanding into auto insurance and home equity financing — HELOCs and mortgages. Sean is clear-eyed that the technological moat in those adjacent categories won’t be identical to what Kin built in homeowners. But it doesn’t need to be. Nine years of customer data — risk profiles, creditworthiness, home characteristics — means Kin enters each adjacent product as a better-informed underwriter than any cold-start competitor, without needing to rebuild the moat from scratch.
The first product created the data asset. The adjacent products monetize it.
Sean’s summary of what drove all of it: “I guess I would say strategy probably in my view trumps execution. I don’t think we’ve always been perfect at executing, but we’re very driven. So usually we recover from the mistake. But I’m 100% sure we have a really excellent strategy.”
Listen to the full conversation with Sean Harper on Unicorn Builders.