Segmentation

Ask a team who their target market is, and you'll get the same answer almost every time.

"Everyone."

Sometimes it wears a disguise. "Anyone who manages projects." "Any business that sends an invoice." Same answer, just more words.

Segmentation is the discipline of picking the specific group of people you'll serve first. Not forever. First. And here's the claim this whole page rests on: build for everyone and you build for no one.

I can hear the objection already. "But the big platforms DO serve everyone. Facebook serves everyone. Airbnb hosts everyone." They do now. Look at how they started and you'll see something different. One specific segment, focused on, dominated, before they touched the next one.

Facebook started with Harvard students. Then other colleges. Then the planet. Airbnb got its start aiming at convention-goers. In 2008 the Democratic National Convention brought more than 84,000 people to Denver, a city already short on hotel rooms, and a scrappy little site called AirBed & Breakfast launched that August to catch the overflow. From there, high-profile events. Then everywhere. And Uber ran the same play. One city, black cars only, a tech-savvy crowd happy to pay a premium, before any of the rest of it.

You have to serve someone before you can serve everyone. Every giant did.

What "everyone" actually costs you

Skipping segmentation isn't a neutral choice. Build for everyone and your discovery, the work of talking to customers and testing your guesses about them, drowns in three specific ways.

Too many options. When your market is everyone, every idea is technically for your market, and you can't tell where to start.

Nothing can be proven. Worse, nothing can be disproven either. "Somebody out there probably wants this" is always true and never useful. So your riskiest assumptions just sit there, untested, quietly running up the budget.

And the feedback comes back as mush. This is the one that kills teams. Run ten interviews drawn from "everyone" and you'll talk to a freelancer, an enterprise IT manager, a retiree, a college student. Their answers will contradict each other, because they should. They have different problems. So you average the feedback, and the average describes nobody. You did everything discovery told you to do and still learned nothing, because the sample was mush before you asked the first question.

Think about the team shooting 50% from the free-throw line in a league where good teams hit 80. The fix isn't "get better at basketball." It's "take our free-throw percentage from 50% to above 75%." Name the one thing and aim at it. Segmentation is that same discipline. Instead of naming the one weakness to fix, you name the one group to serve. One group, not everyone.

The $1,000 bet

Try a thought experiment.

You're forced to bet $1,000 on whether there are any people in the world who want to visit another planet. Before you place the bet, you get to talk to exactly one customer. Who do you pick?

Not a random person off the street. You pick the person with the highest desire to go, because that one conversation tells you the most about whether your bet can possibly pay off.

Your product is the same bet with more zeroes. Your limited time and money only buy you so many conversations, with one small group of customers. So pick the group that values the problem you're solving the most. The rest of this page is one question you keep asking until you can name the room you'd find these people in.

Narrow until you can reach them

The method is one question, asked on repeat:

Among my current group, who would value getting this problem solved the most?

Start at "everyone" and keep asking it. After each cut, run the stop-check: can I easily list physical and digital places where I could strike up a conversation with these people this week? If no, narrow again. If yes, stop.

Let's run it. Say the job you're solving is getting invoices paid on time, the example I keep coming back to in this section. You've built something that helps a business get its unpaid bills collected faster. So who do you talk to first?

From everyone, who values that most? Businesses that invoice, since consumers pay at the register. Can I name a place where I have direct access to "businesses"? Not honestly. Way too broad. Narrow again.

Among businesses that invoice, who feels it most? Small service businesses billing by the project, a two-person design studio, a solo contractor. There's no finance department chasing payments, so every late invoice is a cash-flow problem, not a line item. Can I reach "small service businesses" somewhere specific? Still fuzzy. Narrow again.

Among those, who bleeds the most? Independent consultants and freelancers. One client paying 60 days late isn't an accounting nuisance for them. It's rent. Can I name where to find them? Yes. Two coworking spaces within 15 minutes of me, three online communities where they hang out, a monthly meetup. Stop.

Then prioritize those places by two things: how quickly and cheaply you can get into them (minutes beat hours), and whether you've got a natural excuse to talk to people once you're there.

Notice what the stop-check is doing. It never asks whether the segment is big or impressive. It asks whether you can talk to them. A segment you can't get in touch with is not a segment. It's a daydream. And yes, this can leave you with a group that feels too small. That's fine at this stage. A tiny group you can actually reach this week teaches you more than a huge one you can't.

Fifteen minutes, not an offsite

How long should this take? Fifteen to thirty minutes. Minutes, not days.

"That feels irresponsible. Shouldn't sizing a market take longer than my lunch break?" Sizing a market, sure. That's not what this is. This is picking who to talk to first. And one thing keeps it honest: everything you just produced is still an assumption. No decision rides on it. The segmentation has one job. Get you to a consistent group fast, so your feedback stops being mush. The real conversations test it. So hold the segment loosely. It's your best guess at who to call, not a vow. When the conversations show you picked wrong, and they'll tell you fast, you just narrow again from what you learned.

So don't get theoretical about it. Whiteboarding for a week in search of the perfect segment is just another way of avoiding contact with real customers.

And if the exercise leaves you with more than one reachable segment, pick one. When they tie, let ease of reach break it. The segment you can get in front of soonest wins, because the whole point is to be in real conversations this week. After that, lean toward the one with more profit potential, and after that, toward your own interest or your biggest knowledge gap. Then start.

Saying no to a credit card

Picking a segment has a sharp edge nobody warns you about. It means turning away the people outside it. Including the ones holding money.

When I first opened up my product course, I let anyone sign up. Over time I noticed certain customers used far more of my time than others, and not in valuable ways. They'd monopolize the coaching calls asking questions that made it obvious they hadn't watched a single lesson. So I changed the model. Now there's an application. I interview people who want to join, and I choose who gets to work with me.

Since then, I've turned people down. People who said money wasn't the problem, pulled out their card, and asked me to process it right there. Based on the questions I ask in that interview, and the feeling I get from the answers, I can tell when someone isn't going to put in the work. That's a no. Card or no card.

In discovery, be a filter, not a funnel. A funnel lets everything through. A filter says no on purpose. Segmentation is that same filter pointed at customers, and it costs more, because what you're filtering out is revenue. Yes is the easy word. It's also the expensive one. No is the word that protects the work.

Where product-market fit starts

One more reason the segment matters: it's where product-market fit gets found, if it gets found at all.

Marc Andreessen put the cleanest definition on it in 2007. "Product/market fit means being in a good market with a product that can satisfy that market." A good market means one with real, hungry demand. And you don't find that demand in the abstract. You find it in one specific group that wants this badly. That's the segment.

You don't find fit with everyone. You find it with a segment. Facebook had product-market fit with Harvard students years before it had it with your grandmother. The narrow group is where the signal is strong enough to actually see.

Most of what I've built either got killed before launch, which is exactly what should happen when the fit isn't there, or it launched and died because the fit was never real to begin with. Product-market fit is not a thing you declare in a deck. It's a thing a specific segment either shows you or doesn't. And you can't read that signal at all when your sample is "everyone."

So here's the move. Set a timer for 30 minutes. Write "everyone" at the top of a page, with the job you're solving next to it. Ask the question: who would value this the most? Narrow until you can name the exact places you'd find these people this week. A "place" doesn't have to be a coworking space or a meetup. If your people are scattered inside big companies, the place might be a role on LinkedIn, a Slack community, or three names a colleague can introduce you to. The test is the same: can you actually get one of them on a call this week? Then close the laptop, go to one of those places, and run the conversations.

Because build for everyone and you build for no one. The only way out of that is to pick the someone and go talk to them. Those first conversations will feel awkward and you won't know what to ask. I'll walk you through that craft next, the interview questions that turn a guessed segment into a real one. Talking to customers.