The biggest segment in your market is usually the worst one
Two ponds. One has ten times the fish and everybody is fishing it. The other has fish that jump into the boat. Most teams pick the first, on the grounds that it is bigger, and spend two years learning why that was the wrong reason.
Here is a decision that looks obvious and is not. A software company sells into two platform ecosystems. One of them is enormous and contains most of the category's customers. The other is perhaps a tenth the size. Which do you build your go-to-market around?
Almost everyone answers the big one, and almost nobody runs the analysis that would tell them otherwise. Here is a founder who did, on the larger of his two ponds:
The churn is terrible, the customer support is terrible, there's so much slop competition in their app store, it's just a mess. And half of the people signing up don't really have a business yet.
Jesse Schoberg
He chose the smaller one. The verdict on that choice:
In BigCommerce, the fish are jumping in the boat. And in Shopify, it's probably there, but it's an ocean and we're dragging around. All we see is krill that we don't really want.
Jason Cohen
The number of people who say "most of my customers in e-commerce are Shopify, so we go Shopify" — without the analysis you just did, to say yes, by number, but they're rubbish and I don't want them — that's so smart what you're saying.
Jason Cohen
Why size is the wrong first question
Segment size tells you the ceiling on how many customers could exist. It tells you nothing about the four things that actually decide whether you get any of them.
| What size does not tell you | Why it decides the outcome |
|---|---|
| Whether they can pay | A segment full of people who do not yet have revenue converts at a fraction of the rate and churns at a multiple of it |
| Who else is shouting | Ten competitors in a small pond is a harder market than two in a big one |
| What support costs | Cheap, inexperienced buyers consume more support per pound than anyone else |
| Whether they stay | A big cohort with double the churn is a smaller business than a small cohort that renews |
The instinct to go where the numbers are is really an instinct to avoid a decision. Picking the big pond feels safe because you cannot be accused of missing the market. It is where the largest number of your competitors' emails are also arriving.
The diagnostic that finds the good pond
You do not need a research project. If you already have customers, the answer is in them, and the question is one line:
Where do your happiest, longest-retained customers come from — and what is true about that pond that is not true of the bigger one?
The second half is the part people skip. Knowing that your best customers are recruitment agencies is mildly useful. Knowing that recruitment agencies buy well because they bill their own clients for the output is what lets you go and find another two hundred companies with that property, some of which are not recruitment agencies at all.
If you have no customers yet, the same question runs on the closest available proxy: which of your prospect conversations went somewhere, and what did those companies have in common that the stalled ones didn't?
Scoring two ponds honestly
Score each candidate segment out of five on each line. Do it fast — precision is not the point, and a spread of three points is decisive.
- Ability to pay. Do they have revenue, and is a budget for this kind of thing already normal for them?
- Urgency. Is this a top-three priority for somebody, or a good idea?
- Reachability. Can you get in front of them with the channels you have today, without an ad budget you do not have?
- Competition density. How many vendors are already emailing them about this?
- Retention shape. Does their need repeat, or is it a project that finishes?
- Referral behaviour. Do they talk to each other? A pond where buyers share recommendations compounds; one where they don't means you buy every customer.
The bigger pond usually wins on reachability and loses on everything else. That is the trade you are actually making, and it is much easier to make deliberately than by default.
Two failure modes to name
Averaging. If you serve both ponds, your metrics are a blend of two businesses and describe neither. The blended churn number is the classic case: it is too high to be comfortable and too low to be alarming, so nobody acts, and the good cohort's economics are invisible under the bad cohort's noise. Split the reporting before you argue about the strategy.
Confusing the pond with the fish. "Small business" is not a segment, it is a size band. A pond is a group whose members share a mechanism — the same buying trigger, the same budget line, the same reason your thing pays for itself. If you cannot say the mechanism in a sentence, you have not picked a pond, you have picked a filter.
What to do
- List your last twenty won deals and your last twenty losses side by side, and look for the property that separates them. It is rarely industry and it is often a business-model detail.
- Score two candidate ponds on the six lines above. Ten minutes, out loud, with whoever else sells.
- Split the churn and support numbers by pond before you decide anything. The average is hiding the answer.
- Commit the message, not the company. Choosing a pond means your homepage, your outreach and your next three features speak to it. It does not mean refusing anybody else's money.
Keep reading
- The one number that caps how big your company can get Divide your new revenue each month by your churn rate. That number is the size your company stops growing at — and no marketing channel, however good, can push you past it.
- One channel is no longer a strategy The playbook that said pick a channel and go deep was written for a search engine that judged you on your own pages. Three things changed at once, and the minimum viable channel mix went from one to three.
- Treat the first call as a research interview you also happen to be selling on You will lose most first meetings. The ones you lose are worth something only if you left with an answer you did not have before — which requires deciding, before you dial, what you are trying to learn.