Mobile App Analytics Engineering

Mobile analytics is different from web analytics. Users install, use, and uninstall. Sessions span days, not minutes. Offline usage is common. Privacy regulations restrict tracking. Good mobile analytics tracks what matters — user journeys, retention, and value — without drowning in vanity metrics.

---

Event Taxonomy

# Design a consistent event naming convention
# Format: {category}_{action}_{target}

# User lifecycle events
events = {
    # Onboarding
    "app_opened_first_time": {"source": "install"},
    "onboarding_started": {},
    "onboarding_step_completed": {"step": "int", "step_name": "string"},
    "onboarding_completed": {"duration_seconds": "int"},
    "onboarding_skipped": {"at_step": "int"},
    
    # Core actions
    "search_performed": {"query": "string", "results_count": "int"},
    "item_viewed": {"item_id": "string", "category": "string"},
    "item_added_to_cart": {"item_id": "string", "price": "float"},
    "checkout_started": {"cart_value": "float", "item_count": "int"},
    "purchase_completed": {"order_id": "string", "total": "float", "payment_method": "string"},
    
    # Engagement
    "feature_used": {"feature_name": "string", "context": "string"},
    "notification_received": {"type": "string"},
    "notification_opened": {"type": "string", "delay_seconds": "int"},
    "share_initiated": {"content_type": "string", "platform": "string"},
    
    # Technical
    "error_occurred": {"error_code": "string", "screen": "string"},
    "performance_metric": {"screen": "string", "load_time_ms": "int"},
    "crash_detected": {"exception": "string", "stack_trace": "string"},
}

---

Key Metrics

Acquisition:
  Install rate: Downloads / App Store impressions
  Cost per install: Ad spend / Installs
  Organic vs paid ratio: Organic installs / Total installs
  
Activation:
  Onboarding completion rate: Completed / Started
  Time to first key action: Median time from install to first [purchase/post/etc.]
  Day 1 retention: % of users who return day after install
  
Retention:
  Day 1/7/30 retention: % returning on day N
  Rolling 7-day retention: Active in last 7 days / Active 14-7 days ago
  Cohort retention curves: Retention by install week
  
  Healthy benchmarks:
    Day 1: 25-40%
    Day 7: 10-20%
    Day 30: 5-10%
    (Top apps: 2-3x these numbers)
    
Revenue:
  ARPU: Revenue / Active Users
  ARPPU: Revenue / Paying Users
  Conversion rate: Paying users / Active users
  LTV: Lifetime value per user (predicted revenue over lifetime)
  
Engagement:
  DAU/MAU ratio: Daily active / Monthly active (stickiness)
  Sessions per day: Average sessions per active user
  Session duration: Median session length
  Feature adoption: % of users using feature X

---

Implementation

// iOS Analytics SDK wrapper (Swift)
class AnalyticsManager {
    static let shared = AnalyticsManager()
    
    private var providers: [AnalyticsProvider] = []
    private var userProperties: [String: Any] = [:]
    
    func configure() {
        providers = [
            FirebaseAnalyticsProvider(),
            MixpanelProvider(),
        ]
    }
    
    func identify(userId: String, properties: [String: Any]) {
        userProperties = properties
        providers.forEach { $0.identify(userId: userId, properties: properties) }
    }
    
    func track(_ event: String, properties: [String: Any] = [:]) {
        // Enrich with standard properties
        var enriched = properties
        enriched["platform"] = "ios"
        enriched["app_version"] = Bundle.main.appVersion
        enriched["os_version"] = UIDevice.current.systemVersion
        enriched["timestamp"] = ISO8601DateFormatter().string(from: Date())
        enriched["session_id"] = SessionManager.shared.currentSessionId
        
        // Send to all providers
        providers.forEach { $0.track(event, properties: enriched) }
        
        // Local logging for debugging
        #if DEBUG
        print("📊 Analytics: \(event) \(enriched)")
        #endif
    }
    
    func trackScreen(_ screenName: String) {
        track("screen_viewed", properties: [
            "screen_name": screenName,
            "previous_screen": NavigationTracker.shared.previousScreen ?? "none",
        ])
    }
}

---

Anti-Patterns

Anti-Pattern	Consequence	Fix
Track everything	Data noise, storage costs, analysis paralysis	Track decisions, not clicks
No event naming convention	Inconsistent data, impossible to query	Standardized taxonomy
Client-side only analytics	Lost events, ad blockers, offline gaps	Server-side event validation
Vanity metrics (total downloads)	Misleading growth narrative	Focus on retention and activation
No cohort analysis	Cannot detect if product is improving	Weekly cohort retention curves

Mobile analytics is not about collecting data — it is about collecting the right data, structuring it consistently, and using it to make better product decisions.