Why Customers Stay

Everyone asks why customers leave. It is the wrong question.

I worked this out in a journal entry in February 2022, and I have never found a reason to walk it back: “People don’t leave because they have a reason to leave. And they don’t leave because they are dissatisfied. They leave when they no longer have a compelling reason to stay.” The question is not why do they leave. The question is why do they stay.

That single swap changes everything downstream, because the two questions point in opposite directions in time. “Why do they leave?” sends you digging through the past — the bad ticket, the missed renewal call, the buggy release. “Why do they stay?” forces you to look at the future, which is the only place a retention decision actually gets made.

Here is the proof. Take two customers with the identical past — same onboarding, same outcomes, same complaints — and one renews while the other cancels. The past can’t explain the split, because the past is the same. The only thing that differs is what each one believes is coming next. As I put it at the time: people make different decisions about staying or leaving despite the same past experience, so “the past is relevant to the degree that it informs their belief about their future.”

Customers don’t leave because they have a reason to leave. They leave when they no longer have a compelling reason to stay — and the only compelling reason to stay anywhere is looking forward to better things to come.

So why do they stay? Because they get results — and, more precisely, because they expect to keep getting them. “Renewal is a forward-looking decision. Customers don’t stay because of the results they’ve achieved, but for the results they anticipate.” This is why a product has to always carry a compelling “what’s next” — a result the customer can still reach that they haven’t reached yet. The day the “what’s next” goes quiet, the renewal is already lost; you just don’t see it yet.

This is also where the satisfaction obsession falls apart. Satisfaction is a verdict on the past — how do you feel about what already happened. Retention is a bet on the future. They are not the same axis, which is why happy customers leave and frustrated ones stay. In my own data the connection between satisfaction and retention is effectively zero. What predicts retention is whether customers get measurable results — and customers who achieve them stay six times longer than those who don’t.

Now the deeper layer, which I think is the most important thing I have ever written about churn. Buying and canceling are not judgments about your product at all. “The psychology of sales and churn are the same: a person will purchase, not when he determines the product ‘works,’ but when he determines it is likely [to] work for him. A person will cancel when they determine the product will no longer work for them.” These actions are the outcome of judgments about the self.

I tested it on myself: you can’t get me to buy an expensive home exercise machine by convincing me it works — that it is capable of making a person fit and healthy — because I already believe that. You’d have to convince me that I will actually achieve those results in spite of my own limitations. The conversation is never about the product. It’s about me. Or, stated flat in a 2025 entry: “They already believe it works. The question in their minds is whether it will work for them.”

Buying and canceling aren’t judgments about your product. They’re judgments about the self. Your customer already believes the product works — the only live question in their mind is whether it will work for them.

A subscription, then, is not a transaction. It is the same “will it work for me?” conversation, repeated in perpetuity — and it has to be re-won every cycle, because both the meaning of “works” and the customer’s own capabilities keep changing underneath you. That is the real reason retention can’t be faked: the subscription model keeps pulling you back to the only thing that matters.

If renewal is a forward-looking bet, the obvious next question is what signal a customer reads to place it. The intuitive answer is past results. The intuitive answer is wrong. Past results are an unreliable predictor — “might have been a fluke, temporary, one-time, low-hanging fruit.” The better signal is anything that points to real, permanent change and ongoing effort. Which means the strongest past signal you have is not the result itself — it’s measurement.

The argument that locked this in for me is one line: if past results were the best predictor of future results, nobody would ever change. Improvement is, by definition, a break from past performance. So anyone serious about improving has to ignore their track record and hunt for present, ongoing signals of real change. That is what your customer is doing every renewal cycle, whether they could articulate it or not.

Your best past signal isn’t past results. It’s measurement. Because if past results were the best predictor of future results, nobody would ever change — improvement is the break from the track record.

This reframes the whole job. You’re used to thinking in stated preferences — “how satisfied is the customer?” The signal that actually counts is a revealed preference — “why do customers stay?” — and the longer a customer stays, the more reliable that signal gets. The clearest revealed preference a business has is repeat transactions.

Most retention programs get this backwards: removing reasons to leave does not create a reason to stay. I wrote it in 2017 and it still holds — “the existence of a reason to stay is not undermined by reasons to leave but only by its own disappearance.” You can close every complaint, fix every bug, win back every detractor, and still lose the customer, because none of that is a reason to look forward. Success is not the absence of failure.

There is one more lever worth naming, because it shows up in the data before anything else does: who you sell to. The customers who only sign up with a discount, a free trial, or an opt-out churn at a much higher rate — “the only reason this is not obvious to us is that we don’t measure their churn separately.” In my benchmark, customers who buy without a discount go on to last 2.6 times longer than discount buyers. A reason to buy that isn’t also a reason to stay is bad for growth.

So stop running churn post-mortems. They tell you why the past happened, and the past was never the decision. Ask the only question that points where the decision is actually made: why do they stay? Then go produce the answer — a measurable result they’re getting now, and a compelling one they can still reach next. That is the whole of retention. Everything else is bookkeeping.