Skip to content
MotoMeows

Home Game Data

Game Data

Retention Curves Explained: How to Read the Most Important Chart in Games

What D1, D7 and D30 retention really measure, how cohort curves are built, what healthy and broken curves look like, and the traps that make retention look better than it is.

By Sam Porter Miller · · 8 min read

Retention Curves Explained: How to Read the Most Important Chart in Games: minimal line art of a falling retention curve in orange, teal and blue
8 min left

If you only look at one chart about a game's health, make it the retention curve. It shows how many new players are still around days and weeks after they first started. Revenue, community and word of mouth all grow out of it.

The basic idea

Take every player who started the game on the same day. That group is a cohort. Then count how many of them come back on each following day:

  • D1 retention: the share who return the day after their first session.
  • D7 retention: the share who return on day 7.
  • D30 retention: the share who return on day 30.

Plot those values over time and you get a curve that drops quickly at first and then flattens out.

Two ways to count, very different numbers

There are two common definitions, and mixing them up causes most retention arguments:

DefinitionCounts a player on day N if they...Result
Classic (day N)played exactly on day NLower, stricter numbers
Rolling (day N or later)played on day N or any day afterHigher, flattering numbers

Neither is wrong. Just make sure you know which one a number uses before comparing it with anything else.

Reading the shape

The shape tells you more than any single number.

A steep early drop means the first session didn't land. Players tried the game and didn't find a reason to return. Look at onboarding, the tutorial, load times and the first reward.

A curve that never flattens keeps sliding toward zero. Players who stay a week still leave later. This usually points to a thin mid-game: not enough goals, content or social reasons to continue.

A curve that flattens early is the healthy shape. After the first drop, a core group settles in and keeps playing. The height of that flat part is the size of your real audience.

A curve that rises again after a few days often reflects events, notifications or a weekly reset bringing people back. Worth celebrating, but check it's not one-off.

Comparing cohorts

The real power comes from comparing cohorts: players who started before and after an update, players from different countries, platforms or ad campaigns. If the cohort that started after a new tutorial has a higher D1, the tutorial probably helped.

Keep cohorts large enough. A cohort of 40 players can swing wildly by chance.

Common traps

  1. Mixing old and new players. Retention is about new players. Blending in veterans inflates every number.
  2. Counting reinstalls as new players. Without a stable player ID, one person can appear in several cohorts.
  3. Ignoring the source. Players from a big influencer push behave differently from players who found the game by search. Compare like with like.
  4. Celebrating a single day. Look at the trend across many cohorts, not one good Tuesday.
  5. Comparing with "industry benchmarks" blindly. Genres, platforms and business models differ a lot. Your own past cohorts are the best benchmark.

What to do with it

Use the curve to choose where to work:

  • Weak D1: fix the first session.
  • Weak D7: strengthen the first week of goals and rewards.
  • Weak D30 and beyond: invest in long-term systems, social play and fresh content.

For the bigger picture of what studios measure, read What Is Game Data Analysis?. More breakdowns live in Game Data.

Keep reading