01 · RESULTSBetter onboarding, better money
“After implementing Emhance's recommendations across levels 1–3 of our game, we saw 9.6% more users completing level 3 and as a consequence, 7.3% more users reaching level 10, which resulted in 14.5% LTV uplift.”Marko Petkovic, CEO, Peaksel
02 · MY PARTI owned the study end to end
- Client communication and kick-off: agreed on the question, the decision behind it and the metric that would show success.
- Planning: method, players, scenario and timeline.
- Control: quality checks while sessions ran, not after.
- Insights and presentation: from what players did to why, with recommendations ranked by severity.
- Follow-up: stayed with the client until the change was live and the numbers were in.
03 · QUESTIONWhere was clear. Why was not.
Analytics pointed to the moment before Level 3. It could not show the mechanism behind the exit. The team needed to know whether the puzzles were too hard, too boring, or something else.
We ran first-session playtests with 6 new players, recorded per-second engagement with facial coding, and interviewed each player after the session.
04 · FINDINGA 45-second window of confusion
Players were not blocked by difficulty. They didn't know how to explore. The tutorial built one mental model: guided, linear play on a single face of the vault. The live game then asked for something else:
- non-linear exploration across several faces,
- hidden elements with no prompts,
- several active puzzle zones at once.
Within the first 45 seconds of live gameplay, engagement dropped in every session, and facial coding showed confusion before any real puzzle interaction began.
The core loop itself was fine. When players reached a moment of understanding, engagement rose sharply.
“There’s a sense of delight when the puzzle comes together… it’s a relief.”Player 4
05 · FIXBridge the tutorial and the game
Peaksel adjusted Levels 1–2 so the player's model matched the game before the drop-off point:
- a gradual introduction to exploring several surfaces,
- early exposure to hidden elements,
- a controlled increase in puzzle complexity.
06 · WHY IT WORKEDFrom exit point to exit mechanism
Analytics found the location of churn. Emotional data and interviews found its cause. High engagement during the “aha” moments confirmed that the puzzles worked; the problem was getting players to them.
Public source: emhance.ai, Peaksel case study