In short
- Retention measures whether people keep coming back, which is the clearest sign a product is useful.
- Measure it by cohort: group users by when they started and track each group over time.
- A curve that flattens means a group of people has found lasting value. A curve that falls to zero does not.
- Retention is mostly earned early, through activation, and kept by the product doing its job well.
Also called: Retention, Retention rate, Cohort retention, Churn, Retention curve.
Why retention matters most
Sign-ups show interest. Retention shows value. A product that keeps its users can grow by adding more of them. A product that loses most of its users each month has to replace them constantly, and every marketing dollar leaks out of the bottom.
That is why retention is often treated as the best single signal of product-market fit. People vote with their time. If they come back without being pushed, the product is doing something they need.
How cohort retention works
A cohort is a group of users who started in the same period, such as everyone who signed up in the first week of March. For each cohort, you count how many were still active one week later, two weeks later, and so on. Plotting those numbers gives a retention curve.
Two shapes of retention curve
Falling to zero
- Fewer users remain every week
- No group of people sticks
- Growth depends on constant new sign-ups
- Usually a sign the core value is missing
Flattening out
- Drops early, then levels off
- A steady group keeps using it
- New users add to a growing base
- Usually a sign of real product value
Define active carefully. For most products it should mean doing the core action, rather than opening the app. A calendar tool might count someone as retained if they scheduled something that week. Choose a period that matches how often the product is naturally used: daily for a messaging app, weekly for a work tool, monthly for a budgeting app.
Retention, churn, and revenue
Churn is the other side of retention: the share of users or customers who stop in a period. In a SaaS business it is usually measured on paying customers or revenue, because losing one large account matters more than losing a free user.
| Measure | What it counts | Best used for |
|---|---|---|
| User retention | Users still doing the core action | Product value, early stage |
| Customer churn | Paying customers who cancel | Subscription health |
| Revenue retention | Revenue kept from existing customers, including upgrades | Business health in SaaS |
What improves retention
Retention is decided early. Users who reach real value in their first days are far more likely to stay, so activation and user onboarding are often the biggest levers. After that, retention comes from the product continuing to do its core job well, quickly, and reliably.
- Find what retained users have in common. Which actions did they take early that others did not?
- Talk to people who left. A short call or a one-question email often reveals a fixable reason.
- Fix the core workflow before adding features. New features rarely save a product whose main job is frustrating.
- Build natural reasons to return. Reminders tied to the user's own goals, shared work with teammates, or data that grows more useful over time.
Beware of forced retention
Streaks, spammy notifications, and hard-to-find cancel buttons can raise short-term numbers while making people resent the product. They do not create lasting value.
At Choie, the manifesto says it plainly: if work does not help conversions, retention, or revenue, it is not finished. Ongoing monthly work, from $9,900 a month, takes one product priority at a time, and that priority can be a retention problem as easily as a new feature.
Common mistakes
- Watching total active users. A growing total can hide falling retention when sign-ups are rising.
- Counting any visit as active. Opening a notification is not the same as using the product.
- Using the wrong time period. Measuring a monthly product daily makes healthy retention look broken.
- Waiting too long to look. Retention data starts the day the first user signs up. Track it from launch.
Choie's launch work includes measurement of what people do in the product, so the first cohort can be tracked from the week it arrives.
Common questions
How do you calculate retention rate?
Take a cohort of users who started in the same period, count how many are still active after a set time, and divide by the size of the cohort. Repeat for each period to build a retention curve.
What is the difference between retention and churn?
They describe the same thing from opposite sides. Retention is the share of users who stay over a period, and churn is the share who stop. In subscription businesses, churn is often measured on paying customers or revenue.
What is a good retention rate?
It depends heavily on the type of product and how often it is naturally used, so outside benchmarks can mislead. The clearest signal is a curve that flattens, and one that improves for newer cohorts.
How soon after launch can we measure retention?
You can start measuring immediately, but you need a few cohorts over several periods before the curve means much. For a weekly-use product, that is usually a couple of months.
What should you fix first if retention is low?
Start with the core workflow and the path to first value, since that is where most people are lost. In Choie's ongoing work something meaningful ships every week, which suits retention: change one thing, then compare the next cohort with the last.