Day 1 retention and the 24-hour return rate

Cast

Mallow

SENIOR CONSULTANT · 13Y

ZehnStudio26's senior consultant. Roblox veteran since the early days. Strong at reading the structure underneath the data.

Marsh

ROBLOX BEGINNER · READER VOICE

A brand-side marketer. New to Roblox. Asks "what is that?" so the reader doesn't have to.

Marsh

Retention means Day 1 and Day 7, right? I see those numbers on the Roblox dashboard all the time.

Mallow

Yes, those are the standards. Day 1 Retention is "what share of people who played on day zero also played the next day." Day 7 is "what share are still playing a week in." Both are calendar-day based and ship inside Creator Hub's analytics tab.

Marsh

So if I move those two, that's enough?

Mallow

Mostly. But following the 2026 algorithm conversation, there's a third number that keeps coming up — a 24-hour return rate.

Marsh

How is that different from Day 1? 24 hours is basically the next day.

Mallow

Close, but not the same. Day 1 is calendar-day based — "did the player from Day 0 also play on Day 1." The 24-hour return rate is a rolling window — "did they come back within 24 hours of leaving." That also counts the same-day case: played in the morning, came back in the evening.

Marsh

Ah, so "came back the next day" and "comes back multiple times in a short cycle" are measuring different things.

Mallow

Exactly. Users who habitually re-open the same day will pull the 24-hour return rate up. The benchmark most often cited is 1.2+ sessions per user per day. Below that, the read is "the short-cycle re-engagement loop isn't structurally working."

Marsh

So the retention numbers I should be tracking now are…

Mallow

Roughly: Day 1 and Day 7 — the calendar-day classics on the dashboard. 24-hour return rate — the rolling-window metric in the spotlight, with 1.2 sessions/day as the bar. Day 1 is about next-day return; 24-hour return rate also includes same-day. The design moves are slightly different.

Marsh

Start with next-day, then. What pulls people back tomorrow?

Mallow

For Day 1 Retention, the staples are daily rewards and push notifications. Both are built to fire "open this again in 24 hours" — which is structurally next-day. Roblox ships an Engagement Rewards Feature Package that gives you the UI and data wiring out of the box, so the cost to integrate is low.

Marsh

Daily rewards = log in every day, get coins. That's enough on its own?

Mallow

Weak on its own. Three tweaks. First, always include a streak — day 1 through day 7, ramping the reward so the back half feels expensive to lose. Second, show tomorrow's reward today, so the player has something specific to come back for. Third, sprinkle in a small gacha element for next-day anticipation. The streak break threshold is usually 48 hours, not 24 — if you cut at exactly 24h, a single missed day mass-churns the cohort.

Marsh

What changed on the notification side recently?

Mallow

In 2025, Experience Notifications rate-limiting relaxed from "once every three days" to "once a day." Effectively the recall toolset for devs got noticeably wider. People talk about notification-opt-in users having a 1.5–2× D1 lift — but that varies a lot by genre and design, so treat it as directional.

Mallow

Message framing matters too. Loss-aversion framing ("your streak is about to break," "unclaimed rewards waiting") tends to outperform announcement framing. Since you only get one send per day, sending each user at their personal best hour is the natural operating play.

Marsh

Okay, next up — coming back the same day.

Mallow

The design center of gravity shifts. For the 24-hour return rate — and the 1.2 sessions/user/day benchmark — the core lever is short-round design. A one-hour session won't get opened on busy days. Compress one round to five to ten minutes, and the psychological threshold for "just one more round" collapses.

Marsh

Murder Mystery is like that, right?

Mallow

Murder Mystery 3–7 minutes a round, Tower Defense 5–10 minutes a wave, mini-game collections 2–3 minutes a stage, PvP 5–8 minutes a match. That's the center of the current hit list. On implementation: dropouts at the post-match result screen kill the same-day return, so the standard play is "next match starts in 3 seconds" auto-flow into the next round.

Marsh

What's a time-anchored event?

Mallow

Engineered time slots where "if you miss it, you lose." Morning 7–9 login-bonus 2× window, midday 12–13 lunchtime rare item, evening 19–22 golden-hour event. Hit all three and you've naturally produced three sessions a day. Roblox custom-event measurement became 1-hour-granular in 2025, so verifying the effect is now practical.

Marsh

Running that schedule continuously sounds exhausting on the operations side.

Mallow

Right — so a common play is to run the dense schedule only in launch week. Use launch week to boost the algorithm signal and grab a recommendation slot, then drop back to normal cadence once you're in the retention phase. Two-step operation.

Marsh

Offline progress — the Tycoon staple — which side does that help?

Mallow

This one's interesting — the cap length changes which metric it touches. 24-hour cap trains "return once a day = okay," which leans Day 1 Retention. 8-hour cap nudges three logins a day — morning, midday, evening — which leans 24-hour return rate and 1.2 sessions/day. Same mechanic, different parameter, different metric served.

Marsh

Interesting. What about social mechanics?

Mallow

Social mechanics tend to help both. "Friend just logged in" notifications recover next-day return; Co-Play-only rewards split across midday and evening drive same-day second visits. The 2026 algorithm also reportedly weights a separate signal called 7-Day Intentional Co-Play Days per User, which some say outranks individual D1.

Discussion — Day 1 / Day 7 plus the 24-hour return rate

Where retention metrics stand today

The Roblox Creator Hub dashboard ships Day 1 and Day 7 by default. Day 1 = “share of players from Day 0 who also played on Day 1.” Day 7 = “share still playing a week in.” Both are calendar-day-based classics, and they remain important KPIs. As ballpark genre benchmarks, RPGs tend to land near D1 35%, Tycoons near 30%, Obbys near 18% — wide variation by genre.

Following the 2026 algorithm conversation, a “24-hour return rate” appears to also carry weight. This one is a rolling-window measure — “did the player come back within 24 hours of leaving.” A morning-then-evening same-day return and an overnight return are both counted as a single return event.

The benchmark cited for that rate is 1.2+ sessions per user per day. Users who habitually re-enter the same day raise the rate; below 1.2, the read is “the short-cycle re-engagement loop isn’t structurally working.”

Day 1 and 24-hour return rate are similar but distinct

Easy to conflate, so worth pinning down:

Day 1 captures “did they come back tomorrow.” The 24-hour return rate also captures “did they come back later today.” To move both, you stack the levers that match each.

Plays for next-day return (Day 1 Retention)

Next-day return mechanics are, by nature, “something happens in 24 hours” structures.

• Daily rewards + 7-day streak: Engagement Rewards Feature Package gives you the scaffold
• Push notifications: One per day since the 2025 limit change. Loss-aversion framing (streak about to break, unclaimed rewards) tends to outperform announcements
• Offline progress (24-hour cap): “Come back tonight, the bucket is full” → leans Day 1

Daily rewards and notification opt-in promotion both have a strong cost-to-effect ratio and are usually worth shipping on day one.

Plays for same-day return (24-hour return rate)

Same-day-second-session mechanics need a different mental model.

• Short-round design (5–10 min): The shorter a round, the easier “one more round today” becomes
• Time-anchored events (dense in launch week): Engineer morning/midday/evening “miss-it-and-you-lose” slots. Roblox custom-event measurement is 1-hour granular as of 2025
• Offline progress (8-hour cap): Naturally nudges three logins a day
• Auto-rematching from the result screen: “Next match in 3 seconds” stops the drop

Short-round design is architectural — it’s hard to retrofit. Which is to say it’s worth getting right early.

Mechanics that help both

Some mechanics hit both:

• Social hooks: friend invites, Co-Play rewards, party features, guild missions. Drives next-day and same-day second sessions
• “Friend just logged in” notifications: usable both for next-day and same-day
• The 2026 metric 7-Day Intentional Co-Play Days per User: reportedly outranks individual D1 in the algorithm’s weighting

A unifying lens — “unclaimed reward”

If you step back across these mechanics, the common psychological pattern is “unclaimed reward.” Tomorrow’s daily reward. The full offline-progress bucket. The streak about to break. The event about to start. Each of them leaves something the player could come back for at the moment they close the app.

Stacking several of these so the player always has “a reason to open it again” is, structurally, what we think drives both Day 1 Retention and the 24-hour return rate. That’s the working hypothesis.

The right mix varies by genre

These patterns are what we see in the field, not the only answer. Short-round design fits Murder Mystery and Tower Defense; bolting it onto a setting-heavy RPG can erode the game’s personality. Roleplay-style titles that bank on long sessions may be better served by different KPIs (D7, D30, social touches per stay).

The offline-progress cap is similarly context-dependent. A 3-hour cap on a title with mostly working-adult users punishes them on weekdays and can accelerate churn. Treat the cap as a parameter chosen from your target — Day 1 vs. 24-hour return rate, not as a default value.

Summary — three numbers to hold in mind

If we line them up, the three numbers worth carrying when reading Roblox retention are:

• Day 1 / Day 7: Calendar-day, dashboard-native, classic
• 24-hour return rate: Rolling window, reportedly weighted in the 2026 algorithm
• 1.2 sessions/user/day: The practical benchmark for the rate

Group the plays as “pulls them back tomorrow,” “pulls them back today,” and “social — both.” The numbers above are observational and vary by genre and timing — hold them as hypotheses, validate against your own data. Hopefully this gives you a useful one-perspective frame to work with.