Turning an intuition about learning into a measured onboarding experience—and giving more players a reason to keep playing.
Before-and-after figures describe the supplied two-week windows. Onboarding completion is measured separately from repeat play.
Footpump had launched without an onboarding experience. Players entered a game whose rules took some understanding, and our experience suggested that a little guidance could help them make sense of it.
That gave us a design hypothesis: if players understood the game more clearly, more of them might choose to play again. Before designing the tutorial, we needed a way to see whether that hypothesis held up after launch.
We chose a behavioural measure close to the problem: the distribution of games played per user. Among the people who played Footpump, what share played once, twice, or more?
We measured a two-week window before onboarding launched. 66% of players had played exactly one game, while 15% had played two. The concentration in the first bucket gave us a clear baseline for the work.
Our intended outcome was a lower share of one-game players and a shift toward more games per player. The tutorial itself would also need to be easy to get through, so we treated step conversion as a separate diagnostic measure.
Share of players in the supplied pre-launch window. The complete distribution is compared with the post-launch window below.
We designed the onboarding flow, then tested it through game sessions and interviews. A central prompt was simple:
“Explain the game to us again, in your own words.”
Asking people to describe the rules gave us a way to check what they had understood. We used those conversations and the game tests to revise the flow before releasing the final version.
The design work and measurement plan served the same question: could the introduction help people understand enough to continue playing?
The game had no tutorial. We expected guided learning to make the experience more approachable.
Game tests and interviews checked how players understood and described the game.
We refined the onboarding, then monitored its steps alongside the repeat-play distribution.
The final design moves from right to left. Each measured step has a corresponding event label, connecting the screen sequence to the conversion data.
We tracked the transition into onboarding, the AF1–AF22 sequence, and the gift-pack events at the end. This let us look for individual points where people might struggle or leave, rather than relying only on one completion number.
The final flow teaches inside a practice match. A coach explains the current situation while highlights and hand cues direct attention to the relevant part of the game: the scoreboard, the player’s hand, the opponent’s cards, or the next action.
The sequence builds from selecting and comparing cards to specialty bonuses and energy allocation. Players see those rules alongside the resulting round outcomes. At the end, the gift pack provides a handoff from practice to the main game.
The invitation explains the tutorial and offers a clear starting action.
A coach introduces the practice experience.
The scoreboard is highlighted within the game.
The tutorial points to the player’s cards.
The opponent’s cards are introduced in context.
A hand points to the card selection area.
The selected card is ready for confirmation.
The two selected cards are compared.
The coach explains the comparison.
The result remains visible alongside the explanation.
The scoreboard connects the round result to match progress.
The next card selection continues the practice.
The specialty bonus is shown on the matchup.
The tutorial explains the resulting card power.
The next round result reinforces the example.
Another matchup demonstrates the specialty rule.
Both specialties are visible on the selected cards.
The coach guides the next decision.
The remaining rounds are highlighted.
A slider introduces the energy allocation decision.
The round result follows the energy choice.
The player makes another energy allocation.
The final practice result leads to the reward.
A gift pack marks the end of the practice sequence.
The opened pack reveals the reward cards.
Of 19,606 users in the initial pool, 17,470 pressed Start: 89.11% entered the tutorial. That first transition accounted for a 10.89% drop, the largest in the flow.
Our interpretation was that some people had clicked into Footpump out of curiosity and were not ready to begin a tutorial. The funnel identifies where they left; that explanation remains the team’s interpretation of the behaviour.
After Start, every measured transition converted between 97.36% and 99.93% of users from the preceding step. There was no later drop of a similar size. The largest later declines were about 2.64% at AF15 and 2.63% at AF2—useful places to keep watching as the experience evolved.
Users reaching each event ÷ users at the previous event. The first point uses all 19,606 users as its denominator.
Scroll sideways to explore all 25 events.
The CSV’s Percentage column is cumulative against all users. This chart recalculates the previous-step rate from UserCount. The final event represents 14,526 users: 74.09% of all users and 83.15% of the 17,470 who started.
| Event | Users | From previous | Drop-off | Of all users |
|---|---|---|---|---|
| V1_OnboardingBtnStart | 17,470 | 89.11% | 10.89% | 89.11% |
| V1_OnboardingAF1 | 17,458 | 99.93% | 0.07% | 89.04% |
| V1_OnboardingAF2 | 16,999 | 97.37% | 2.63% | 86.70% |
| V1_OnboardingAF3 | 16,883 | 99.32% | 0.68% | 86.11% |
| V1_OnboardingAF4 | 16,763 | 99.29% | 0.71% | 85.50% |
| V1_OnboardingAF5 | 16,656 | 99.36% | 0.64% | 84.95% |
| V1_OnboardingAF6 | 16,496 | 99.04% | 0.96% | 84.14% |
| V1_OnboardingAF7 | 16,457 | 99.76% | 0.24% | 83.94% |
| V1_OnboardingAF8 | 16,424 | 99.80% | 0.20% | 83.77% |
| V1_OnboardingAF9 | 16,373 | 99.69% | 0.31% | 83.51% |
| V1_OnboardingAF10 | 16,316 | 99.65% | 0.35% | 83.22% |
| V1_OnboardingAF11 | 16,258 | 99.64% | 0.36% | 82.92% |
| V1_OnboardingAF12 | 16,106 | 99.07% | 0.93% | 82.15% |
| V1_OnboardingAF13 | 16,053 | 99.67% | 0.33% | 81.88% |
| V1_OnboardingAF14 | 15,966 | 99.46% | 0.54% | 81.43% |
| V1_OnboardingAF15 | 15,545 | 97.36% | 2.64% | 79.29% |
| V1_OnboardingAF16 | 15,478 | 99.57% | 0.43% | 78.95% |
| V1_OnboardingAF17 | 15,457 | 99.86% | 0.14% | 78.84% |
| V1_OnboardingAF18 | 15,416 | 99.73% | 0.27% | 78.63% |
| V1_OnboardingAF19 | 15,376 | 99.74% | 0.26% | 78.42% |
| V1_OnboardingAF20 | 15,258 | 99.23% | 0.77% | 77.82% |
| V1_OnboardingAF21 | 14,942 | 97.93% | 2.07% | 76.21% |
| V1_OnboardingAF22 | 14,638 | 97.97% | 2.03% | 74.66% |
| V1_OnboardingGiftPack | 14,609 | 99.80% | 0.20% | 74.51% |
| V1_OnboardingGiftPackOpened | 14,526 | 99.43% | 0.57% | 74.09% |
Initial pool: All Users = 19,606 (100%). Counts and cumulative percentages: Each step.csv. Previous-step conversion and drop-off are calculated from those counts.
We measured the same games-per-user distribution during a two-week window after launch. The share who played exactly one game fell from 66% to 43%, a reduction of 23 percentage points.
The share who played twice rose from 15% to 18%, and the share who played three times rose from 7% to 10%. The broader distribution also moved toward more games per player.
Share of players by number of games played, measured over two weeks before launch and two weeks after launch.
Values are the whole-percent figures supplied in the CSVs, without normalization. They total 97% before and 95% after; 0% is a reported rounded value, not proof of no users. ≈ marks the approximate 11+ totals, calculated by summing rounded rows for 11 or more games. Cohort sizes and exact dates were not supplied.
| Games played | Before | After |
|---|---|---|
| 1 | 66% | 43% |
| 2 | 15% | 18% |
| 3 | 7% | 10% |
| 4 | 3% | 5% |
| 5 | 2% | 4% |
| 6 | 1% | 3% |
| 7 | 1% | 2% |
| 8 | 1% | 1% |
| 9 | 0% | 0% |
| 10 | 1% | 2% |
| 11 | 0% | 1% |
| 12 | 0% | 1% |
| 13 | 0% | 1% |
| 14 | 0% | 1% |
| 15 | 0% | 0% |
| 16 | 0% | 0% |
| 17 | 0% | 1% |
| 18 | 0% | 1% |
| 19 | 0% | 1% |
| 20 | 0% | 0% |
| 21 | 0% | 0% |
| 22 | 0% | 0% |
| 23 | 0% | 0% |
| 24 | 0% | 0% |
| 25 | 0% | 0% |
| 26 | — | 0% |
| 27 | 0% | 0% |
| 28 | 0% | 0% |
| 29 | 0% | — |
| 30 | — | 0% |
| 31 | 0% | 0% |
| 32 | 0% | 0% |
| 33 | 0% | — |
| 34 | 0% | 0% |
| 35 | 0% | — |
| 36 | 0% | 0% |
| 37 | 0% | 0% |
| 38 | 0% | 0% |
| 39 | — | 0% |
| 40 | 0% | 0% |
| 41 | 0% | 0% |
| 42 | 0% | — |
| 43 | 0% | — |
| 44 | 0% | — |
| 46 | — | 0% |
| 47 | 0% | — |
| 48 | — | 0% |
| 49 | 0% | — |
| 52 | — | 0% |
| 53 | — | 0% |
| 54 | 0% | — |
| 55 | 0% | 0% |
| 56 | 0% | — |
| 57 | — | 0% |
| 58 | — | 0% |
| 59 | — | 0% |
| 61 | — | 0% |
| 62 | — | 0% |
| 63 | — | 0% |
| 65 | — | 0% |
| 67 | — | 0% |
| 70 | 0% | — |
| 74 | — | 0% |
| 81 | 0% | — |
| 83 | 0% | — |
| 87 | 0% | — |
| 89 | 0% | — |
| 92 | 0% | — |
| 93 | 0% | — |
| 100 | — | 0% |
| 101 | 0% | — |
| 102 | 0% | — |
| 105 | — | 0% |
| 110 | — | 0% |
| 113 | 0% | — |
| 122 | — | 0% |
| 124 | — | 0% |
| 142 | — | 0% |
| 149 | — | 0% |
| 158 | 0% | — |
— means the game-count row is absent from that source file; it has not been replaced with zero. All rows, including the long tail, are preserved here. Sources: Before_Onboarding.csv and After_Onboarding.csv.
The post-launch results moved in the direction we had defined as success: fewer people stopped after one game, and more appeared in the repeat-play buckets. Alongside the high conversion through the teaching steps, this gave us a positive signal for the onboarding experience.
The useful lesson was to define the behavioural outcome before designing. Game tests helped us refine the learning experience; step conversion helped us locate friction; the before-and-after distribution helped us judge whether the overall behaviour improved.
This was a before-and-after comparison, so it does not isolate onboarding as the sole cause of the change. It does show that the measured outcome improved after launch—and gives us a concrete basis for the next iteration.
Measurement notes: before and after each cover two weeks, as reported by the project author. The supplied distributions contain rounded percentages, with no cohort counts or exact dates. No statistical-significance or long-term-retention claim is made. Onboarding event counts come from a separate funnel export.