Your next customer is probably already buying

The customer who becomes your next market is often already buying from you—showing up as an odd order, a review describing an unexpected use case, or traffic no campaign explains. I’ve spent 15 years looking for signals like these inside consumer brands. Spotting the anomaly is easy; identifying the people behind it and deciding whether they justify investment is the hard part.

At MOST, we call this process Customer Diagnostics. I recently discussed it with Andy Walsh on Startups Decoded. Andy’s workbook helps you assemble a list of overlooked customer groups. What follows is the next step: deciding which groups are genuine market opportunities.

Why established customer segments crowd out new growth

Optimizing the customer you already have feels responsible, and looking for one you cannot yet name feels like a hobby. So teams pour budget into the audience they can describe — which is nearly always the audience their largest competitor defined first.

That competitor has more money than you. Bidding for the same people means winning only the ones they did not want. As I told Andy:

If you try to copy your biggest competitor, you’re not just overpaying for their table scraps. You’re making them more money too.

The deeper cost is not the media spend. It is that a team busy defending a known audience never builds the habit of asking who else has this problem — and that question is the only one that produces a genuinely new customer.

How overlooked customer segments appear in your data

Like a mistake. Overlooked customers arrive as anomalies, and anomalies get explained away, because every reporting system is built around the customer the company already decided it has.

When I was CMO of Greater Than, the brand sold an $8-a-bottle sports hydration drink positioned for athletes. Then we started seeing traffic spikes we could not explain and customer comments filled with photographs of breast milk.

Breastfeeding mothers were reporting that the product helped with fatigue and low supply. We had never marketed to them. They had found it — and some were tagging ten friends at a time inside private groups. The team leaned in. Three years later, a roughly $300,000-a-year business was doing $10 million annually, profitably.

The signal that started it was only about 15–20 unusual comments. That is the uncomfortable lesson: the thing that changed the company was small enough to dismiss, and it looked like noise right up until it didn’t.

How do you validate an overlooked customer segment?

This is the question the workbook stops short of, and it is the one that actually costs money to get wrong. Most teams that go looking for overlooked customers come back with a list of eight plausible groups and no way to rank them.

At MOST we triage them with four questions, in this order. A group that fails the first two is not worth testing regardless of how large it looks.

1. Do they have the same problem, or just resemble your buyers? Strip the category language and write down what your product actually does. A sports drink is an electrolyte solution with no added sugar. The mothers were not athletes; they had the same underlying problem. Groups that merely look like your customers are a segmentation exercise. Groups with the same underlying job are a market.

2. Did they arrive without being invited? Unprompted arrival is the strongest signal available, because it proves demand exists without spend behind it. A group that responded to your campaign is telling you about your marketing. A group that showed up on its own is telling you about the world.

3. Do they talk to each other? The mothers at Greater Than were tagging ten friends each in private groups. That is not a nice detail, it is the entire economics: a group with dense internal networks compounds, while a diffuse group of equal size needs paid reach for every single customer. This is where reachable and unreachable get decided, and unreachable is a real answer worth having early.

4. Can you serve them without changing the product? If the answer requires a new formulation, a new SKU, or a regulatory path, it is a roadmap item, not an experiment. Keep those separate. Confusing them is how a discovery project becomes an eighteen-month build.

Signal	Reads as noise	Reads as a market
Where it came from	You ran a campaign at them	They arrived unexplained
The problem	Similar demographics	Same underlying job
Group structure	Diffuse, no shared spaces	Dense, already talking
What serving them takes	New product	New message

This is the first pass behind a Growth Scan: turn unexplained demand into a ranked set of customer-acquisition opportunities before committing a larger budget.

Why more acquisition spend doesn’t solve customer discovery

The failure is not one of resources, it is one of measurement. Sophisticated marketing teams — teams with real budget and good people — still underinvest in discovery, and the reason is usually sitting in their attribution window.

Most windows are built to reward things that pay back in seven days. Anything whose payoff is slower gets starved before it can prove itself, and discovery is by definition slower: you are looking for a customer nobody has taught to buy yet. Find your product’s real latency — the average time between first touch and first purchase — and measure against that instead. If you do not know that number, it is this week’s job.

How to run a low-cost customer discovery test

What makes an experiment worth running is not confidence. The goal of experimentation isn’t to be right — it’s to make being wrong cheap.

That reframes the whole exercise. You are not trying to pick the correct group out of your list; you are trying to design a test whose downside you can absorb without a meeting. For example, if being wrong costs five hundred dollars and a week, you can afford to be wrong four times and still be ahead of the team that spent a quarter building certainty about one option.

Fifteen years of running this process by hand is what MOST is built on. The bottleneck is doing it rigorously and repeatedly across customer feedback, acquisition data, competitors, and market signals. MOST applies the same triage across those sources, ranks the opportunities, and turns the strongest candidates into bounded tests.

So: who is the customer you might be overlooking? The answer is probably in your last twenty support tickets.