Game Analysis

Level Churn Diagnosis

Pinpoint sticking-point levels, analyze the churn root cause, and output optimization recommendations, for all level-based progression games.

IndustriesCardRPGSLGMatch-3Runner

Agent · Skills Live

skill · loadedindustry+100

Run this Skill on my data

Conversion funnel

stage to stage

100%

74%

52%

31%

Analysis readyevidence attached

Level Churn Diagnosis is a level-health diagnosis capability in ThinkingAI's industry Skill library. It helps game operations teams identify the sticking-point levels causing user churn, analyze the churn root cause, and provide actionable optimization recommendations, turning "which level is driving users away" from vague intuition into precise data. This is ThinkingAI's decade of industry experience, packaged into ready-to-use Skills.

Live walkthrough

AgentSave as AgentExport session

level-churn-diagnosis

Over the last 30 days, new-user 3-day retention dropped from 35% to 28%. I suspect some level is driving people away, but with 60-plus levels we can't check them one by one. Help me pinpoint it.

Pull the last-30-day new-user "last attempted level" distribution to locate the churn concentration point

Compute the clear rate and average attempts for levels 1–40

Compare against the historical baseline (same definition 60 days ago) to rule out a global decline

Analyze behavioral traits of the blocking level (attempts / post-failure return rate)

Compare level mechanic configurations

Level 25 has a gameplay-mechanic switch

Found it, it's level 25. Your retention drop can basically be explained by this one level.

1. Blocking-point localization

Among new users who churned in the last 30 days, 25.9% last stopped at level 25: the runner-up (level 12) is only 7.3%. This is a classic single-point block, not a global loosening.

Level 20–30 clear rate (new users)　Unit: %

2. Behavioral traits of level 25

Metric	Level 25	Adjacent levels average
Clear rate	42%	78%
Average attempts	6.8	2.1
Next-day return rate after failure	31%	67%

3. Mechanic-level root cause

Comparing level configs reveals: level 25 is a gameplay-mechanic switch point. The first 24 levels are all "timed match", and level 25 switches for the first time to "move-limited". New users hit the rule change with no transition, fail 6–7 times in a row, and quit. This isn't a pure numeric-difficulty problem, it's a design-type block.

Heads-up

Working back from the retention target: if level 25's clear rate recovers to 60%+, then by the churn-share estimate, 3-day retention could rebound about 4–5pp. Recommend a segment validation first, to confirm whether it's "new users can't adapt" or "the level itself is overtuned".

Look at it by segment: if veteran players also can't pass, then it's the level itself; if only new users get stuck, the onboarding guidance may be the problem.

Segment definition: new users (registered ≤7 days) vs veterans (registered 30+ days)

Compare the two groups' level 25 clear rate / attempts

Cross with the "move-based gameplay tutorial" completion rate

Generate an optimization recommendation list

The segment results support your second hypothesis: it's mainly a new-user adaptation problem, and the guidance step does have a gap, though the difficulty is also on the high side. Two factors stack:

Metric	New users	Veterans
Level 25 clear rate	35%	61%
Average attempts	7.4	3.2
Move-based tutorial completion rate	44%	N/A (veterans' version had no such tutorial)

Two signals: (1) veterans' clear rate is 61%, passable but still below the level average of 78%, which says the level itself is on the hard side; (2) new users' move-tutorial completion is only 44% (other guidance steps are all above 90%), so the tutorial itself is skipped or interrupted, and more than half of new users are brute-forcing it without understanding the new rules.

Optimization recommendations (by priority)

1.Make the move tutorial mandatory (currently skippable) and move it forward to the level 24 results page

2.Raise level 25's move limit 25 → 30 to pull first-clear difficulty back within the curve

3.On first failure, show a mechanic hint + failure compensation (+3-move item) to break the quit-after-consecutive-losses pattern

4.Move the mechanic switch point later or add 1–2 transition levels (to assess next version)

Conclusion

Items 1–3 can be done via hotfix and should ship this week; item 4 involves level-sequence changes, so put it in the next version. After the changes ship, I'll automatically track the recovery curve of level 25's clear rate and 3-day retention.

Follow-up data two weeks later: level 25 clear rate 42% → 58%, new-user 3-day retention rebounded to 33%.

On your data

That was a simulated run

Leave your work email and we will run a live walkthrough on your real business data.

The problem

Churn in level-based progression games often concentrates in a few sticking-point levels: 30% of churned users concentrate in 1 or 2 sticking-point levels, whose churn rate can be more than 5x that of normal levels. Operations teams usually only know overall retention is down, not which specific level is driving users away. The traditional approach of checking retention and clear rate level by level means pulling data 30 times for 30 levels, and a full investigation takes 1 to 2 days.

What it does

Automatically identify sticking-point levels: analyze the last level users attempted, locate the churn concentration, and output a churn-rate and clear-rate comparison table

Root-cause analysis down to the level: distinguish difficulty-type, fatigue-type, and design-type sticking points

Segment comparison: performance differences between new and existing users, and paying versus free users, on the same level, to inform differentiated interventions

Before-and-after version comparison: quantify a new version's or event's impact on level churn

Output targeted optimization recommendations: concrete plans for difficulty tuning, reward design, guidance mechanics, and social boosts

When to use it

A level-based game (match-3, card, RPG) with declining overall retention that needs the churn sticking point located

Assessing whether a revision improved churn after a level's difficulty was adjusted

A large performance gap between new and existing users on the same level that needs a differentiated difficulty strategy

A clear clear-rate gap between paying and free users that needs the paid-level design assessed

Comparing before-and-after level churn after a major version update or holiday event

In the field

Case

A match-3 game · level 25 sticking-point diagnosis

Overall 3-day retention fell from 35% to 28% over the past 30 days. The Skill found level 25 was the biggest sticking point: 25.9% of churned users last stopped at that level, with a clear rate of only 42%, far below the 78% average of the first 24 levels. Further analysis found the mechanic abruptly changed from timed clearing to a move limit, with a new-user clear rate of only 35%. It recommended loosening the move limit and adding a first-failure hint and failure-reward compensation.

FAQ

Does it only apply to match-3 games?

No. It applies to any game with a level-progression mechanic, such as card stage-clearing, RPG dungeons, SLG city advancement, and runner tracks.

How is churn defined?

The default is 3 days inactive, adjustable by game type: 1 to 2 days for light casual and 5 to 7 days for heavy RPG.

What is the difference between a sticking-point level and funnel churn?

Sticking-point diagnosis focuses on quitting caused by in-level difficulty design, while funnel churn analysis focuses on the conversion rate of the overall flow.