Game Data Analytics System Integration
Firebase Analytics in Unity connects in half hour. But month into production when Product Manager asks "why purchase conversion 2x lower on Android vs iOS"—turns out all events log correctly, but parameters partially lost exceeding 25 custom parameters per event limit. Retention funnel not set up at all because session_start and first_open auto-counted, but level completion never sent.
Analytics is not "connect SDK". This is designed event model agreed with game design and business metrics.
Where Data Gets Lost
Wrong event taxonomy. Firebase limits: 500 unique event names per app, names to 40 chars, parameters to 25 per event, string values to 100 chars. Projects logging level_complete_world_1_level_5_stars_3 as event name quickly hit limit and lose historical data flexibility.
Right schema: level_complete as event, world_id, level_id, stars, time_sec, attempts as parameters. Allows slicing in BigQuery without changing client code.
GA4 sampling. Firebase Analytics (GA4) applies sampling at large data volumes in standard interface. For accurate data need BigQuery Export—free in Firebase, but requires setup. Without BigQuery, funnels on millions show approximate figures.
Event duplication. In Unity with DontDestroyOnLoad analytics manager, after new scene load old instance not destroyed—events sent twice. FirebaseAnalytics.LogEvent() not idempotent, dupes invisible in raw stream without group by on BigQuery side.
What Integration Includes
Design event model—jointly with game designer and client analyst. Determine: which events needed for retention analysis (D1/D7/D30), monetization funnel, difficulty balance, A/B tests. Record in event taxonomy document before coding.
SDK connection: Firebase Analytics as base, plus if needed—Amplitude (better for behavioral analysis), GameAnalytics (free, good for mobile hyper-casual), or AppsFlyer/Adjust for attribution. Each needs separate init with GDPR/ATT.
Implementation: abstract IAnalyticsService over concrete SDKs—switch providers without changing game code. AnalyticsManager—singleton via ServiceLocator, not MonoBehaviour—removes scene lifecycle dependency.
Setup BigQuery Export from Firebase Console, create basic analytics queries for reporting (retention, funnel, revenue by segments), optionally—dashboard in Looker Studio.
Remote Config as Part of Analytics Cycle
Firebase Remote Config with A/B Testing not just collects data but tests hypotheses. Typical: change level 3 difficulty for 10% audience, measure D1 retention and purchase conversion difference. Requires proper user grouping by FirebaseRemoteConfig.FetchAndActivateAsync() and config applied before controlled screen display.
Timeline
| Scale | Timeline |
|---|---|
| Firebase Analytics, basic event model, one platform | 3–7 days |
| Firebase + BigQuery + attribution SDK | 2–3 weeks |
| Complete stack with Remote Config, A/B, dashboards | 4–6 weeks |
Cost calculated after analyzing metric requirements and current analytics infrastructure state.