Betting on what you know: lessons from landing in a new market

A few weeks ago we had to decide what to bring to San Francisco. Not metaphorically: we were about to land in a new market and we needed something in our hands when we got there.

The alternative was showing up empty, and I wasn’t sure which option scared me more.

Showing up with something

The dilemma is always the same. Arrive with something you’re not sure about, or arrive with nothing and try to figure it out from zero while the clock is running. Neither is comfortable. One of them usually buys you conversations. The other buys you silence.

Arriving empty sounds humble, but in practice it means you don’t have a starting point, and the people you want to learn from don’t have anything concrete to react to. Abstract conversations are the most expensive kind. Specific ones move faster, even when what you’re showing turns out to be wrong.

So we decided to show up with something.

The bet

We already had a WhatsApp bot live in Italy. A mature one, with real usage: booking management, review collection, a handful of agentic flows that had been debugged against real restaurant owners for months. The product worked, the playbook worked, the infrastructure around it worked.

The natural move, heading into the new market, was to lean on that. Same channel, same core product, same team intuition. A rational bet given what we knew.

But “rational” and “verified” are not the same word, and I knew that going in. We were making a decision with partial information because waiting for complete information was more expensive than the risk of being wrong.

A bet is not a mistake. A bet is a decision you make when waiting is more expensive than being wrong. That’s not a justification, it’s the shape of the problem.

The ground truth

We landed in San Francisco and started doing the thing you can only do on the ground: talking to restaurant owners. Real ones, one at a time, in their places, at the times they were willing to give us.

It didn’t take many conversations to learn something that’s obvious in hindsight. They don’t use WhatsApp. Not the way Italy does. Not for business. Not for this.

The product hypothesis held up fine. The channel hypothesis didn’t. Two different hypotheses, two different outcomes, and only the second one was wrong. That’s a useful thing to learn on day three instead of month six.

There was no drama in the moment. Just new information, which is what I was there for.

What a bet is actually worth

Here’s the part I had to sit with, because it’s less intuitive than it sounds.

Without a working product in hand, we wouldn’t have had a reason to be at those tables. We wouldn’t have had something specific to show, and the owners wouldn’t have had something specific to react against. The conversations that corrected our direction happened because we’d made a bet, not in spite of it.

The bet gets you into the room. The room tells you what to build next.

A bet that turns out wrong isn’t a failure of judgement. It’s the cheapest way to buy feedback from people who wouldn’t have given you their time in the abstract. The only way to truly waste one is to cling to it after the room tells you it’s wrong.

And remote research is not field research. What people actually do, what they actually prefer, what channel their real exchanges happen on, often doesn’t show up in data you can google. It shows up in the tenth sentence of a conversation, the one you’d never have triggered without being there.

Now we research properly

So now we’re doing the part we couldn’t do from Italy. Figuring out what channel restaurants in San Francisco actually use to handle bookings, reviews, and customer exchanges. Which of our flows translate, which need to be rebuilt around a different medium, and which simply don’t apply. The work isn’t blocked, it’s redirected.

Research isn’t a phase you finish before building. It’s a discipline you keep running in parallel to everything else. The moment you stop, your preconceptions start making decisions for you silently, and silent decisions are the worst kind.

The bet got us to the conversations. The conversations are now rewriting the plan. That’s the loop.

If you’re about to bet

If you’re about to walk into a new market, a new audience, or a new problem, you’ll have some version of this choice. Bring the thing you already know how to build, or bring nothing and try to learn from scratch.

Bring the thing. Just bring it with two honest rules attached.

Name your assumption out loud. Not in your head, on paper. “We’re betting that X is the right channel / customer / pricing / workflow.” If you can’t write it down cleanly, you don’t understand your own bet well enough yet.

Go find the evidence that destroys it, not the evidence that confirms it. The goal of the first conversations isn’t to validate your bet. It’s to pressure-test it against reality and come back with sharper questions.

A good bet doesn’t pretend to be truth. It’s a hypothesis dressed up as a starting point, which is all any of us really have when we’re moving into something new.

The bet gets you to day one. The research is how you get to day two.