Anyone can show that store users stick around. World of Carrom set out to prove something far harder, that the store itself was the reason. Here’s how a strict digital-twin study answered it, and why the answer reaches all the way down to ARPDAU.
Inside World of Carrom 3D, one of India’s most-installed carrom titles, PlaySuper moved the single number that decides a mobile game’s year: first-week retention. Across 2,248 behaviourally identical players, those who engaged with the PlaySuper rewards store retained 3.6 percentage points better at Day 7 than those who didn’t, opened the game 10.5% more often, and were active on 9.8% more days. Since that lift lands in the earliest, most valuable stretch of a player’s life, it doesn’t stay put, instead it compounds downstream into ad impressions, IAP conversion and ARPDAU.
The more impressive part is what those numbers survived to be stated. There’s a comfortable lie that inflates half the case studies in mobile gaming: players who use our feature retain better, therefore our feature drives retention. It ignores that the people who opt into anything, such as a store, an event, or a season pass were never a random sample; they were already the engaged ones, the ones who’d have stuck around anyway.
So when AppOn, the Pune studio behind World of Carrom, wanted to know whether its PlaySuper store was actually keeping players or merely being used by the players who’d stay regardless, it refused the comfortable answer and ran the uncomfortable one instead.
The game, and the stakes
World of Carrom 3D is one of India’s most-installed carrom titles. The game has crossed 50 million downloads on Google Play, holds a 4.1-star rating, and has 100,000+ verified reviews.
At such a scale, a game isn’t fighting for novelty; it’s fighting for the first week of a player’s life. World of Carrom competes for the same Day 7 attention as larger Ludo and casino-style titles, where retention curves are notoriously steep and the first week determines the next twelve months of revenue. Win Day 7 and you own a year, lose it and the install was a rounding error.
The intervention under the microscope was narrow. PlaySuper runs an in-game rewards store; new players who engage with it inside their first two weeks receive curated, high-intent rewards tied to how they play. The hypothesis was equally narrow: does that store engagement produce a real lift in early retention once you strip out selection bias?
Building behavioural twins
The method is our whole story here, because it’s what makes the numbers believable. Rather than compare store users to the general population — the comfortable lie — PlaySuper Analytics built what it calls a strict digital-twin study.
Every player in the treatment group (those who engaged with the store between Days 0 and 13) was paired with a single control player who had not. However, the pairing wasn’t loose. Each pair was closely matched across five baseline behavioural dimensions, all captured before any store interaction could have coloured the results: the player’s discovery week, their game opens in Days 0–2, their coins earned in Days 0–2, their coins burned in Days 0–2, and their transactions in Days 0–2. In other words, same lifecycle stage, same starting engagement, same economic behaviour, same spending, and same activity intensity.
The result of that matching is the reason to trust everything downstream. After pairing, the standardized differencebetween the two groups on every observable baseline behaviour sat below 0.04, a statistical threshold indicating strong balance between the groups. The study ran across a 14-day window from 3 to 16 April 2026, and its sample tightened as the horizon lengthened: 6,883 treatment against 6,369 control at Day 1, 5,915 against 4,981 at Day 3, and a strict 1:1 of 2,248 matched pairs at Day 7, the primary endpoint.
The headline: +3.6 points where it counts
At Day 7, PlaySuper engagers retained at 62.5%, against 58.9% for their behavioural twins—a gap of 3.6 percentage points.
On its own, 3.6 points might read as modest. But the +3.6pp lift is 2.4 times higher than the full-period average lift of +1.5pp, suggesting the effect is not spread evenly across the player lifecycle but is concentrated in the early days, when each retained player has the greatest potential value. Day 7, the study argues, is the inflection point where a player either becomes a durable player or churns for good. D7 retention is widely used as an early indicator of longer-term engagement and lifetime value. A lift there does not sit still—it cascades.
A curve, not a spike
A single data point can flatter. The more persuasive evidence is the shape of the full early-retention curve, which moves in one direction throughout the measurement period. At Day 1, store users retained at 37.0%, versus 35.7% for their matched players—a +1.3pp difference. By Day 3, the gap widened to +2.5pp (53.0% vs. 50.5%). By Day 7, it reached +3.6pp (62.5% vs. 58.9%).
Taken together, the pattern is clear: the retention lift roughly doubles from Day 1 to Day 3 and widens again by Day 7. Rather than following the pattern of a short-lived novelty effect, where initial curiosity fades after the first session, the results suggest a sustained improvement in early player engagement. Players who interacted with the store did not simply show a brief spike in activity; they showed stronger retention throughout the first week.
The leading indicator underneath
Retention is a lagging metric; by the time it can be measured, the behaviour driving it has already happened. The forward signal is engagement intensity—how often players return and across how many distinct days—and both metrics moved in line with the retention results.
PlaySuper engagers opened the game 3.6 times per player, compared with 3.3 for their behavioural twins: a +10.5% lift. They were active on 2.1 distinct days, compared with 1.9 for the control group: a +9.8% lift.
The second number matters more than it may initially appear. Two players with the same total session count are not equally likely to return if one concentrates all activity on Day 1 while the other spreads sessions across the week. According to the study, day spread is a strong behavioural predictor of D14 and D30 retention, and PlaySuper increased it by roughly 10%. A denser, more distributed engagement footprint is therefore more difficult to churn from.
Why retention is really a revenue story
Here’s where the study makes its most important move: it argues that a retention lift is never just a retention lift. Every monetisation event in a free-to-play game requires a live session. No session, no impression; no impression, no IAP touchpoint; no touchpoint, no payer. Retention is the gate every downstream revenue event must pass through, which is why a lift at the gate can multiply through everything behind it.
On ads, the logic remains mechanical. Ad revenue scales with impressions, and impressions scale with surviving DAU. Model the two survival curves forward, and a retention curve sitting ~3.6pp higher can produce a comparable proportional uplift in cumulative ad inventory across the cohort’s lifecycle. Every additional retained day creates ad-break and rewarded-video opportunities that otherwise would not have existed.
On in-app purchases, the effect is non-linear. First-purchase probability is not flat across a player’s lifecycle. Drawing on public benchmarks for casual board and skill titles—figures the study identifies as illustrative ranges—it remains very low in Days 1–2 (around 0.1%), rises through Days 3–6 (roughly 0.3–0.6%), peaks in Days 7–14 (around 1.0–1.5%), and then plateaus.
The point holds regardless of the precise numbers: players who churn before Day 7 never reach the high-conversion window. Because PlaySuper’s lift is concentrated in the D3-to-D7 period, it moves more engaged players into and through the peak IAP window. This may yield incremental payers without changing the offer or pricing—simply because more of the right players remain active long enough to receive the opportunity.
Stacked together, those effects land on ARPDAU. The study presents a deliberately directional decomposition, indexing the behavioural-twin baseline at 100: a retained-DAU advantage adds roughly +8.5, session depth another +6.0, and the richer IAP conversion window a further +5.5, bringing the PlaySuper cohort to approximately 120.
This is explicitly an illustrative model, not a directly measured revenue outcome. It is built on effects with direct mechanical support in the study data and excludes harder-to-isolate factors such as referrals, social features, and event monetisation. In the study’s words, it represents “a floor, not a ceiling.”
What AppOn does with it
The findings point to three concrete actions.
First, push store exposure into the Day 0–2 onboarding flow. The retention lift is strongest early, and players who engage during their first two days are more likely to benefit from the full Day 7 trajectory.
Second, treat PlaySuper-engaged Day 3 retainers as a priority audience for premium IAP offers. By Day 3, they already outperform their matched counterparts across engagement metrics, making the D3–D7 window a potentially efficient period for conversion-focused offers.
Third, re-run the analysis at Day 30 to test whether the advantage compounds over time rather than decays—an important assumption behind the study’s monetisation implications.
The study’s conclusion is straightforward: across 2,248 closely matched behavioural twins, the PlaySuper store did not merely attract players who were already likely to stay. The analysis suggests that it was associated with stronger early player retention.
Curious what a retention lift like this would be worth in your own title? PlaySuper will model the upside for your game at [email protected].
Read the full World of Carrom × PlaySuper case study here: World of Carrom x Playsuper.
