The Grandmother Test

We know people churn. We don't know who, when, or what it's worth

Three numbers make churn actionable, and an average tenure is none of them: which segment (with its count), when in the lifecycle the risk actually peaks, and what a saved customer is worth in margin rather than revenue. Without the third, you cannot tell whether saving them is worth the effort.

Gergana Tyaneva · 21 September 2026 · 12 years in product and marketing analytics

The grandmother version

A leaking bucket. You can weigh the water that's left every morning and write the number on a chart, and you will have a very accurate record of a bucket emptying.

What you actually need is: which hole, how big, and is it worth the cost of the patch.

A hole near the top leaks only when the bucket is full. One near the bottom leaks always. Same bucket, same total loss, completely different jobs.

Averages hide the hole

"Average customer lifetime is 7 months" describes nobody. It is usually a mix of a large group who leave in week two and a small group who stay for years, and the two need opposite interventions.

Report the distribution: percentiles, buckets, and where the concentration sits. Then cut by segment — and always carry the count, because a dramatic churn rate on n=23 is a story about 23 people.

the average (5.4%)the peak — week 3this is where an intervention has leverageweek 0week 11% of those still hereAverage tenure hides the peak. The peak is the only part you can act on.

When, not how long

The useful question is not average tenure but where the risk spikes. Plot the hazard: of the people who reached week 3, what share left during week 3?

Most consumer products have one or two clear peaks — often right after onboarding, and again at the first renewal. That peak is where an intervention has leverage; spending the same money in a flat period does very little.

Worth what, exactly

Value a save in margin, not revenue. A rescued customer on a heavily discounted plan can be worth a fraction of what the headline number suggests.

And decide your "gone" threshold before you measure, not after. For most consumer products around two weeks of silence is a reasonable line. State the threshold you used — it changes the answer, so hiding it makes the answer unfalsifiable.

The short version

Who, when, and worth what. Three numbers, each with its count attached.

Get those and churn stops being a number on a slide and becomes a decision about where to spend the next three weeks.

Churn, retention and LTV by source are the heart of Where the Money Comes From — €4,500.

See prices →

More answers

Attribution — Which channel actually pays?Plumbing — Our numbers live in eight tools and no two agreeTime — Month-end reporting eats three days and nobody reads itThe real one — Someone asked me for a dashboard and I can't work out what for

One question before you read on. I'd like to count visits with Google Analytics, which sets cookies and sends your IP to Google. Nothing has loaded yet, and the site works exactly the same either way. What's collected.