Intent guide

Mobile Game Success Criteria

Do not start with a D1 scoreboard. Start with whether a stranger can finish one loop, then read return and stability only after that gate clears - and send each bad signal to one lesson, not five new features.

Success is a triage, not a vanity KPI

Retention dashboards are useful after the product is understandable. If strangers cannot complete the loop, those numbers mostly measure confusion.

Hold a short pre-launch bar and a short post-launch bar. Promote a build only when the pre-launch bar is honest.

Ask different questions before and after launch

Pre-launch asks whether the product can be understood and survived. Post-launch asks whether the same intent earns another sitting and whether systems still leave players choosable.

Do not use post-launch metrics to excuse a broken first session.

Android vitals overviewCrash-free and ANR-free rates are public store-quality signals, not optional polish.
Play Console · Android vitalsUse the official vitals surface when judging technical readiness to trust behavior data.

Pre-launch criteria

Understandability: a new player can name what to do next without a coach
First session: notice → try → succeed → want-again happens early
Control readability: inputs map to outcomes players can explain
Core-loop completion: the four beats close without a feature tour
Performance and crash risk: a full sitting survives on the target device class

Post-launch criteria

Return signal: players come back for the same kind of intent, not only a push
Session behavior: sittings still end choosable, not flooded or starved
Progression blockers: players can see why they are stuck
Technical stability: crash and ANR rates stay in a shippable band for your store
Review quality: written feedback names the loop, not only price or ads

Stop at the first No

Walk the tree top-down. Stop at the first No and open the matching lesson instead of adding another dashboard tile.

Measurement decision tree

Ask the left lane in order. The right lane names what you may trust only after that gate clears - a no answer means fix the lesson area first.

Ask in order

Understand the loop?
Can a stranger complete one loop without help? If no: stop KPI hunting and fix teaching first.
Same-loop return?
Do returning players re-enter the same loop? If no: check fairness and attributable outcomes.
Stable enough to trust?
Is the technical baseline quiet enough to trust behavior data? If no: stabilize runtime ownership first.

Only then trust

First-session want-again
Measure whether strangers choose another try of the same loop.
Session and progression signals
Inspect session budget leftover and whether progression still shapes the anchor decision.
Leading signals for the next lesson
Use quiet behavior data to pick the next lesson instead of chasing vanity KPIs.

Understand the loop?Same-loop return?then ask
Same-loop return?Stable enough to trust?then ask
Understand the loop?First-session want-againif yes
Same-loop return?Session and progression signalsif yes
Stable enough to trust?Leading signals for the next lessonif yes

Ask the left lane in order. The right lane names what you may trust only after that gate clears - a no answer means fix the lesson area first.

Reading order

Understand the loop?
Can a stranger complete one loop without help? If no: stop KPI hunting and fix teaching first.
Same-loop return?
Do returning players re-enter the same loop? If no: check fairness and attributable outcomes.
Stable enough to trust?
Is the technical baseline quiet enough to trust behavior data? If no: stabilize runtime ownership first.
First-session want-again
Measure whether strangers choose another try of the same loop.
Session and progression signals
Inspect session budget leftover and whether progression still shapes the anchor decision.
Leading signals for the next lesson
Use quiet behavior data to pick the next lesson instead of chasing vanity KPIs.

Connections

Understand the loop? → Same-loop return? (then ask)
Return metrics are meaningless if strangers never complete the loop once.
Same-loop return? → Stable enough to trust? (then ask)
Behavior data is only trustworthy when crash and ownership noise is low enough to read.
Understand the loop? → First-session want-again (if yes)
Only after a stranger can finish one loop should you measure first-session want-again.
Same-loop return? → Session and progression signals (if yes)
Same-loop return unlocks reading session leftover and progression shape as design signals.
Stable enough to trust? → Leading signals for the next lesson (if yes)
Leading signals guide the next lesson only when the runtime baseline is quiet enough to trust.

Leading signals pick the next lesson

Leading signals tell you what to fix this week. Lagging signals tell you whether last month’s fix held. Prefer leading signals while the loop is still learning.

Play pre-launch reportPre-launch technical scans complement, and do not replace, first-session understandability checks.

Leading and lagging signals

Signal	Kind	Phase	Meaning
Time to first successful loop	leading	pre-launch	Shows whether teaching works before you scale traffic.
Unprompted re-entry in session	leading	pre-launch	Players choose another try of the same kind without a forced gate.
Crash-free sittings on target devices	leading	pre-launch	Behavior data is trustworthy only if the session survives.
D1 return to the same activity	lagging	post-launch	Confirms the loop earned another day - not only a notification open.
Session leftover readability	leading	post-launch	Sittings end with a next decision, not a silent pile of currency.
Store review themes about fairness or confusion	lagging	post-launch	Language in reviews often names attribution failures before charts do.

Common failed diagnoses

These illustrative misreads show how a healthy-looking number can hide the wrong problem. Each pairs with a better next check.

Failed diagnoses

Chasing D7 while first success is late

Mistake: Optimizing retention creatives and economy knobs before strangers can finish one loop.

Better: Fix first-session teaching and feedback timing, then remeasure return.

Reading spend as engagement

Mistake: Treating a flooded soft wallet as proof players love the session.

Better: Check session budget and whether leftover points at a next choice.

Blaming ads for fairness complaints

Mistake: Assuming negative reviews are only monetization when players cannot see cause and effect.

Better: Trace whether outcomes were visible before the next decision.

Bad signal → one lesson

This is the product of the guide: a triage board into the learning graph. One bad signal should send you to one deep lesson, not five new features.

Bad signal → lesson

First session never completes the loop
Open Teach the loop in the first session: Rehearse notice → try → succeed → want-again before adding systems.
Players act but cannot see what changed
Open Check the feedback loop: Make the consequence reach the next decision.
Outcomes feel arbitrary
Open Fairness perception: Fairness is often attribution, not spreadsheet balance.
Sittings end flooded or starved
Open Design a one-session economy budget: Budget earn and spend for one sitting before week-scale tuning.
Reward shape feels flat or spiky
Open Progression pacing curves: Read the curve silhouette across the session, not only the total.
Resources lack a decision purpose
Open Game economy flow: from goal to feedback: Trace goal → choice → signal before adding more currencies.

Continue into the graph

These lessons deepen the decisions on this page. Each link includes why it matters next.

Lesson link map

Teach the loop in the first session
First-session teaching failures show up before any retention chart matters. (~6 min)
Check the feedback loop
Invisible outcomes poison both fairness and return intent. (~7 min)
Fairness perception
Review language often names attribution before metrics do. (~7 min)
Design a one-session economy budget
Flooded or starved sittings distort every economy KPI. (~6 min)
Energy as a session gate
Gates that cut mid-loop create churn that looks like a retention bug. (~6 min)
Progression pacing curves
Flat or spiky reward shape changes whether the next sitting feels worth it. (~6 min)
Game economy flow: from goal to feedback
Resources need a decision purpose before you trust spend charts. (~6 min)
Ship Checklist: From Vertical Slice to Store Submission
Crash paths and store rejections distort success metrics as badly as a weak loop. (~7 min)

Game Design foundations

When understandability and fairness are the risk.

Economy foundations

When session leftover and progression shape are the risk.

Browse all topics · All guides