Your Game Has Players, But They Don’t Come Back
Retention is a symptom with several possible causes. Find which one you have before you build anything.
Your concurrent players look fine on a good day, yet the people who arrived yesterday are not there today. The usual response is to add content. Sometimes that is right. Far more often the players never reached the content that already exists.
What D1 and D7 actually measure
- D1
- The share of a cohort that comes back the day after first playing. It mostly reflects the first session: did it make sense, did it reward them, did anything invite them back.
- D7
- The share that returns a week later. It reflects whether a longer loop exists: progression, goals, friends, events.
- Session length
- How long a player stays in one visit. It says nothing by itself about whether they return.
Check how your dashboard defines a returning day before comparing numbers. Compare against your own history and against games with a similar audience, not against a figure from a forum post.
Session length and return rate are separate problems
| Pattern | A likely reading | Where to look first |
|---|---|---|
| Short sessions, weak D1 | Players leave before the game makes sense or pays off | The first five minutes |
| Long sessions, weak D1 | The session was fine but nothing pulls them back, or they ran out of things to do | The end of the session and the reasons to return |
| Fair D1, falling D7 | The early loop works and the middle of the game stalls | Pacing around the first progression wall |
| Good in some cohorts, bad in others | The audience differs by source, device or country | Retention split by cohort |
Find where players disappear
- Log first-session steps: joined, first input, first reward, first time through the core loop, five minutes in.
- Find the largest drop between two adjacent steps.
- Split that funnel by acquisition source, device and new versus returning players.
- Watch real sessions. Make a fresh account yourself, then sit with someone who has never seen the game and say nothing.
- Write down one hypothesis for the leak and the smallest change that would test it.
Onboarding or core loop?
If most players leave before completing the core loop once, you have an onboarding or clarity problem. If they complete it several times and still leave, the loop itself lacks variety, progression or a reason to continue. More content only helps the second case, and only for players who get that far.
Why added content rarely fixes retention
Content sits downstream of the leak. If most new players never leave the starting area, a new late-game zone is seen by almost nobody. It also changes the game while you are trying to measure it, which makes the next diagnosis harder.
Measure this before changing anything
- Weekly retention cohorts for the last two months, so you can see whether the problem is new or old.
- The first-session funnel described above.
- The spread of session lengths, not only the average. An average hides a pile of 40-second visits.
- Retention by traffic source.
- What returners did in their first session that non-returners did not. This comparison is often the most useful of the five, because it points at the behaviour worth encouraging.
For the full funnel view, read A Better Way to Diagnose a Roblox Game’s Growth.