Crash Analytics

# Crash Analytics

You help triage, prioritize, and reduce app crashes — and understand how crash rate affects App Store discoverability and ratings.

## Why Crash Rate Is an ASO Signal

- **App Store ranking** — Apple's algorithm penalizes apps with high crash rates
- **App Store featuring** — High crash rate disqualifies editorial consideration
- **Ratings** — Crashes are the #1 cause of 1-star reviews
- **Retention** — A crash in the first session destroys Day 1 retention

**Target:** crash-free sessions > 99.5% | crash-free users > 99%

## Tools

| Tool | What it provides | Setup |
|------|-----------------|-------|
| **Firebase Crashlytics** | Real-time crashes, ANRs, symbolicated stack traces | Add `FirebaseCrashlytics` pod/SPM package |
| **App Store Connect** | Crash rate trend, crashes per session | Built-in, no code needed |
| **Xcode Organizer** | Aggregated crash logs from TestFlight + App Store | Xcode → Window → Organizer → Crashes |
| **MetricKit** | On-device diagnostics, hang rate, launch time | iOS 13+, automatic |

**Recommended:** Crashlytics (real-time alerts + search) + App Store Connect (trend validation)

## Crashlytics Setup

### iOS (Swift)

```swift
// AppDelegate or @main App struct
import FirebaseCore
import FirebaseCrashlytics

@main
struct MyApp: App {
    init() {
        FirebaseApp.configure()
        // Crashlytics is auto-initialized
    }
}
```

### Non-fatal errors (track without crashing)

```swift
// Log a non-fatal error
Crashlytics.crashlytics().record(error: error)

// Log a custom key for debugging context
Crashlytics.crashlytics().setCustomValue(userId, forKey: "user_id")
Crashlytics.crashlytics().setCustomValue(screenName, forKey: "current_screen")
```

### Android (Kotlin)

```kotlin
// build.gradle (app)
implementation("com.google.firebase:firebase-crashlytics:18.x.x")

// No additional code needed — auto-captures unhandled exceptions
// For non-fatal:
FirebaseCrashlytics.getInstance().recordException(throwable)
```

## Triage Framework

Not all crashes are equal. Prioritize by impact:

**Priority Score = Crash Frequency × Affected Users × User Segment Weight**

| Priority | Criteria | Response time |
|----------|---------|---------------|
| P0 — Critical | Crashes on launch / checkout / core feature; >1% of sessions | Fix today |
| P1 — High | Crashes in common flows; >0.1% of sessions | Fix this release |
| P2 — Medium | Edge case crashes; <0.1% of sessions | Fix next release |
| P3 — Low | Rare, non-blocking crashes; <0.01% of sessions | Backlog |

### Crashlytics Dashboard Triage

1. Sort by **"Impact"** (unique users affected), not frequency
2. Group: `onboarding`, `checkout`, `core feature`, `background`, `launch`
3. Assign P0/P1 to the top 3–5 issues
4. Set a **velocity alert** in Crashlytics for any issue affecting >0.5% of users

## Reading a Crash Report

```
Fatal Exception: com.example.NullPointerException
  at com.example.UserProfileVC.loadData:87
  at com.example.HomeVC.viewDidLoad:45

Keys:
  user_id: 12345
  current_screen: "home"
  app_version: "2.3.1"
  os_version: "iOS 17.3"
```

**Steps to debug:**
1. Open the file and line in Xcode (`UserProfileVC.swift:87`)
2. Check what can be nil at that point
3. Reproduce with the user context (OS version, device, screen)
4. Write a failing test before fixing

## Symbolication

Crashlytics auto-symbolicates if you upload dSYMs. If you see unsymbolicated traces: