AppMetrica vs Firebase Analytics: migrating mobile app analytics

Firebase in Russia has been patchy since 2022, and by 2026 the situation has not improved. Push via FCM still limps along, but Analytics, Remote Config, and Crashlytics drop out periodically: events fail to arrive, the app does not receive remote config and ships with fallbacks. For apps living mostly on the Russian market, keeping Firebase Analytics is a gamble. The closest replacement in features and maturity is AppMetrica from Yandex.

What you get in AppMetrica

AppMetrica covers about 90% of Firebase Analytics + Crashlytics:

• Parameterised events, funnels, retention, cohort analysis
• Crash reports (Android, iOS), including native and ANR
• Push notifications (separate module, works through FCM/APNs externally and through its own channel for Russian devices without Google Play services)
• Remote config (relatively new, available since 2024, less flexible than Firebase)
• Install attribution and in-app purchases
• Free for most projects: limit of 3000 events per user per day, which covers 95% of apps

What is worse:

• BigQuery-style raw event export — present but more limited than Firebase. Data freshness lags up to 24 hours
• ML-based "Predictions" (Firebase) — not available. If you relied on them, build your own or pick a third party
• A/B testing — a separate Yandex Cloud product, less convenient than Firebase A/B

Alternatives to AppMetrica

If AppMetrica is not a fit:

• Amplitude — strong behavioural analytics, excellent funnels and cohorts. Works from Russia, paid (from $49/month above the 100K MTU free tier)
• Mixpanel — Amplitude-equivalent. Also accessible, also paid
• PostHog — open source, self-hostable. Covers analytics + session replay + feature flags + A/B. Great if you have ops capacity
• Devtodev — game and mobile product analytics, Russian, integrates with ad networks

For most "normal" business apps (delivery, marketplace, media, banking), AppMetrica covers the needs. For apps with deep cohort analytics and dedicated BI teams, Amplitude or PostHog are stronger.

Migration plan from Firebase to AppMetrica

Typical timeline is 2-4 weeks per app. Stages:

1. Event audit. Export every event currently sent to Firebase. Not "according to the code," but actually arriving — Firebase Analytics has a report. Split into "need" and "legacy, skip"
2. Schema mapping. Firebase events are name + params. AppMetrica is the same, but parameter names and formats differ (for example, currency in purchases). Build a mapping table
3. Dual emission. The critical step. Add the AppMetrica SDK alongside Firebase and send events to both. At least one month. During this period reconcile numbers: 95% match is normal, lower means something is off
4. Parallel reports. Marketers and PMs get used to AppMetrica, comparing numbers to their old Firebase dashboards
5. Turn off Firebase Analytics. About 6-8 weeks after dual emission starts. You can keep the Firebase SDK only for FCM if push goes through it

Rule of thumb: do not rip out Firebase until marketing confirms AppMetrica covers their workflows.

What we migrated and where we tripped

Over the past six months we migrated three apps. Sharp edges, by severity:

• Event parameters. Firebase tolerates spaces and special characters in params; AppMetrica is stricter. Staging looked fine, production lost half the events. Solution: a schema linter before send
• Attribution. Firebase attribution uses Dynamic Links + Google Play install referrer. AppMetrica uses its own system and Yandex Direct. If you advertise on Google Ads, you will need to switch tracking to AppMetrica deeplinks
• Custom user properties. Firebase lets you set "plan = premium" and filter cohorts. AppMetrica has an equivalent through app profile, with a different syntax. Do not forget to port it
• Crashlytics. AppMetrica has its own crash system, different from Crashlytics. Stack traces decode automatically, dSYM/ProGuard symbols are uploaded via CLI. A couple of days to set up plus a month to accumulate statistics

How much the migration costs

For one app with 50-100 event types:

• Audit — 1-2 days
• Adding AppMetrica SDK (Android + iOS) — 1-2 days
• Dual emission and mapping — 3-5 days
• Testing and reconciliation — 2-4 weeks calendar, 2-3 days of effort
• Porting funnels, dashboards, team training — 2-3 days

Total 10-15 engineer-days + 4-6 calendar weeks. Best with a single focused person — context switching eats half the schedule.

What not to do

• Do not switch Firebase off immediately — you lose the "truth" you reconcile against
• Do not port every event — half were legacy as of 2022. Use the migration as a cleanup
• Do not ignore sessions. AppMetrica counts a session differently (default timeout 30 seconds vs Firebase's 30 minutes). That shifts "events per session" and "time in app" significantly
• Do not leave Firebase Analytics on "just in case" after migration — it burns battery and data on users' phones

Should you migrate at all

If your app:

• Runs mostly in Russia/CIS → yes, migrate now
• Has 50%+ Russian users and a meaningful share of Android without Google Play services (Huawei, ROMs, alt stores) → yes, migrate
• Global with up to 20% Russian share → wait, keep both SDKs for redundancy
• No Russian presence at all → do not migrate, stay on Firebase

Unlike the situations driven by MAU caps or expensive exports, the Firebase problem in Russia is not financial — it is reliability. When data is missing for half a day, marketers do not know how much they are spending on ads, and that is no longer "we will reconcile later" — it is a live fire. AppMetrica wins simply by running consistently inside the local network.