Your churn rate is not a reason

There is a specific kind of meeting that happens in every SaaS company. Churn is up forty basis points. Everyone agrees this is bad. Someone pulls up a cohort chart. Someone else suggests it might be seasonal. Forty minutes later the meeting ends, and nobody in the room has learned a single thing about why any individual human being decided to stop paying.

This isn't a failure of analytics. Analytics did its job perfectly — it told you the number, segmented it, and plotted it against last quarter. The problem is that the number was never going to contain the answer, and no amount of slicing will make it.

Rate is an outcome. Reason is a cause.

A churn rate is the sum of a large number of individual decisions, each made by a person with specific reasons at a specific moment. Averaging them produces something real but useless: a measurement of the outcome, containing none of the causes.

The uncomfortable part is that you can improve at measuring indefinitely without getting any closer to a decision. A dashboard that refreshes hourly instead of daily tells you the same nothing, faster.

The reason has a shelf life

Here's what makes this genuinely hard rather than merely neglected: the reason exists, in a recoverable form, for about a week.

On the day someone cancels, they can tell you precisely what happened. Which feature was missing. Which competitor's demo they watched. The specific Tuesday it stopped being worth the money. Ask a month later and you get a vague, reconstructed, socially acceptable version — 'it just wasn't the right fit' — which is true and completely useless.

So the window in which the most valuable data in your company can be collected is also the window in which you are least likely to think about collecting it, because someone just cancelled and you're annoyed.

Why surveys don't fill the gap

The obvious answer is an exit survey, and the obvious answer mostly doesn't work — for two reasons that compound.

• Almost nobody fills them in. You're asking someone who has just decided to stop giving you money to click a link, load a page, and complete a form. The activation energy is enormous and the reward is zero.
• The people who do respond aren't representative. Filling in an exit survey requires a strong feeling. Mild disengagement — the single most common reason people leave — produces no strong feeling at all, so it's systematically absent from your data.
• Fixed options force people into your existing guesses. A dropdown can only return an answer you already thought of, which means it structurally cannot tell you the thing you don't know.

You end up with a small, biased, pre-categorised sample and a quiet, false sense that you've measured something.

The second question

Even when you do get a free-text answer, the first one is almost never the real one. Not because people lie, but because first answers are short by default.

Filed as-is, that's a price complaint, and the obvious response is a discount. Ask one more question — was there a moment it stopped feeling worth it? — and you might get this:

That is not a price problem. It's a packaging problem, and the fix is a viewer seat, not a discount. Same customer, same opening answer, completely different roadmap decision — and the entire difference is one follow-up question that most companies never ask because there's nobody available to ask it.

What to actually do

None of this requires a tool. It requires that three things happen reliably, and reliability is the hard part:

• Every cancelled customer gets asked, within hours, in a way that costs them nothing to answer. Not a link — a question they can reply to.
• Every answer gets one follow-up, written from what they actually said rather than from a script.
• Every conversation ends up somewhere the whole set can be read at once, because six people saying the same thing across eight months is invisible in an inbox.

Do that by hand for a quarter and you will learn more about your product than your dashboard has told you all year. Most founders manage it for about two weeks — which is precisely the gap WhyLeft exists to fill.