Retention

Retention is a metric that shows how loyal a product’s users are. You take a group of people who opened an app for the first time on the same day. Then you count how many of them are still around after a week, after a month, or after a year. If, out of 1,000 new sign-ups, 200 remain active after 30 days, the retention rate is 20 percent. The counterpart is called churn: the share of those who drop off. Both figures describe the same phenomenon from two sides.

Why investors look at the retention curve first

You can buy new users. Anyone who spends enough money on advertising will almost always get installs and sign-ups. But whether a product is actually good is only revealed by retention. It is hard to manipulate, because no one can be paid to voluntarily open an app every week.

For a company, retention is also directly tied to money. A streaming service earns many times more from a subscriber who stays for three years than from a customer who cancels after two months. The acquisition costs are the same in both cases. That’s why a business model can fail purely due to poor retention, even if the influx of new users looks impressive.

A common mistake is equating high user numbers with success. If a platform grows by two million users in a quarter but simultaneously loses 1.9 million, that’s not growth — it’s a bucket with a hole in it. Experts call this a leaky bucket. Only once the hole is closed does advertising spend have a lasting effect.

Cohorts, curves, and the point where it flattens out

Measurement is usually done in cohorts. A cohort is a group of users with the same start date, for example everyone who signed up in March. This group is tracked over weeks, and the remaining active users are plotted on a curve. Cohorts prevent fresh users from diluting the numbers of older users.

Such curves always drop steeply at the beginning. Most people try something out and quickly lose interest. What matters is whether the curve flattens out afterward or keeps declining toward zero. A flat trajectory indicates a stable core of regular users, and this core is the actual foundation of the business.

The definition of “active” also matters. For a messaging app, active might mean daily use, while for a tax app it might mean just once a year. Companies choose this definition themselves, and generous definitions make the numbers look better. Anyone reading quarterly reports should therefore always check exactly what was counted.

Retention in chatbots, subscriptions, and games

Retention and churn appear regularly in tech companies' quarterly figures. Netflix and Spotify report on cancellation rates, mobile carriers on contract switching. In gaming, people talk about Day 1, Day 7, and Day 30 retention — that is, the share of players who return after one, seven, or thirty days. Values above 40 percent on the first day are considered good there.

For AI products too, this metric has become the most important litmus test. Many chatbots and image generators had enormous sign-up numbers but lost the majority of their users within just a few weeks. Anyone who tries an application only once out of curiosity generates no lasting revenue. Analysts therefore no longer compare just downloads, but the return rates of individual providers.

In everyday life, one encounters the topic from the other side. Discounts offered when you try to cancel, reminder emails, or series with cliffhangers are all retention measures. They aren’t automatically unfair, but they show how much effort a company puts into making sure you stay.