From $15M to $1.8M and Back: How Dover Found PMF Twice (George Carollo)

Hi everyone and welcome to Before It Clicked. Today we go deep with George Corolo, co‑founder of Dover, on the rare journey of finding product market fit twice. Dover’s now a lean, profitable business around six million a year, built on a marketplace for fractional recruiters, but it once rode Autopilot to fifteen million top line before the 2022 hiring freeze knocked the floor out. Buckle up; it’s bumpy and instructive.

We’d just jumped from roughly twelve to fifteen and I thought, wow, this might really be working. Then big tech layoffs hit, rates rose, the market flipped, and the hiring problem changed overnight.

Set the stage. Where’s Dover now and what do you actually do?

We’re a small, scrappy team doing about six million in revenue, flirting with profitability, and we run a marketplace that embeds fractional recruiters into startups. It’s been a windy path to get here.

How did you find your first fit with Autopilot?

I joined as the third co‑founder despite not wanting to do recruiting again. We started as hands‑on recruiters for a few friends’ companies with no product at all. Everyone kept saying the same thing: I need more and better candidates. We leaned into sourcing, built targeted outreach and matching, and gradually automated the workflows. That became Autopilot.

Crowded space, no clear why now. Why believe you’d find something?

The market’s massive and the pain’s perennial. We took a leap of faith. No grand why now—just a belief that a hairy, fragmented problem leaves room to build.

What were the goals and early funding moves?

We cared about building something durable more than quick wins. We entered YC after pivoting away from a totally different idea, raised around twenty‑three million total, and turned our manual work into software. One caution: starting as a service confuses who’s accountable—your product or your people.

How fast did Autopilot grow, and how did it unravel?

We raced from one to three to six to fifteen. Then the 2022 pullback hit. Candidates flooded the market, recruiters were laid off, and our product suddenly competed with an in‑house hire. NDR had just looked great; then demand shut off like a switch. With roughly four hundred logos, eighty‑plus people, and burn around two hundred thousand a month, we had to react.

What did those first months look like?

We waited a bit, then cut about a third of the team and tried to reframe everything. We tested self‑serve, changed pricing to per‑candidate, reworked the funnel—six to nine months of heavy iteration—and it didn’t land. We were tweaking a problem the market had moved past. So we built a free ATS in 2023 to capture mindshare, knowing others charged real money for it. It got to about a thousand companies, but it didn’t fix the revenue hole, and Autopilot kept bleeding.

How did internal dynamics shape those decisions?

We had no outside board seat, so it was mostly the three of us. Change is hard when managers fear for their teams and playbooks. We value open pushback, but at some point you have to move or the company stalls.

So where did the marketplace spark come from, and what was the why now?

A friend asked if I knew anyone who needed recruiting help. That nudge made it click: founders ultimately want a person they trust to own outcomes, and sourcing tools churn because there’s no lock‑in. The why now was clear: AI makes teams more leveraged, funding is tighter, hiring is spiky, and full‑time recruiters are harder to justify. Fractional help fits this reality. We aim to be a high‑end, professional version of Upwork, not a pay‑per‑placement agency with misaligned incentives.

How did you test it, and what told you there was a second PMF?

I ran a solo experiment for a few months, matched the first recruiter with companies, and got instant referrals on both sides. By the fourth customer, the organic pull showed up. Meanwhile the team kept improving the ATS. Later we asked marketplace recruiters if they wanted Autopilot; they passed. That’s when we shut it down and deleted years of code. My view is most sourcing tools churn hard unless you own the data.

Convincing the team to bet the company on the marketplace—what did that take?

I built a case with early metrics and unit economics, but we only had a few months of data, so LTV questions loomed. It took around six weeks to get buy‑in across co‑founders and managers, and we ended up changing leadership. Going from fifteen million top line to under two was brutal; the stress was very real. In the moment, tiny tweaks feel safer than a bold jump, but the ship was sinking.

Any lessons on when to keep iterating versus making a sharp turn?

You only hear the stories when slogging pays off, not when it doesn’t. We asked ourselves constantly if we were iterating toward nowhere. In our case, carving out founder time to run clean, fast experiments was key. Some failed; one broke through.

Where are you taking Dover now?

We doubled revenue last year and want to do it again. We’re polishing the marketplace, smoothing the MVP edges, and continuing to invest in the ATS because it feeds the brand and the ecosystem.

Any asks for listeners?

We’re always bringing great recruiting partners onto the platform to work with founders. That’s our focus right now.

I hope this reminds founders to treat PMF as a living thing. Get it, protect it, and be ready to pivot if it slips. Rapid growth comes with a cost if the market turns, but you can fight your way back. Thanks for the candor—this was eye‑opening.

Thank you. I call that stretch the lost era. It’s strange to re‑enter it with thirty people depending on you. It’s much harder than being lost with three. Stay paranoid about keeping product market fit.