instrumentation-planning

Instrumentation Planning

Strategic guidance for what to measure using Jobs-to-be-Done framework.

Core Question

For each user job, ask:

• Did it complete? → completion rate
• How long? → duration (p50, p95, p99)
• What failed? → error type, context
• Did they give up? → drop-off rate

Priority Tiers

Tier	Focus	Examples
1	Crashes & errors	Crash reporting, error boundaries
2	User context	User ID, session ID, device info
3	Performance	App start, screen load, network
4	Breadcrumbs	Navigation, user actions
5	Business metrics	Funnels, feature usage

Implementation Order

Day 1:  Crash reporting + User context
Week 1: Breadcrumbs + Release tracking
Week 2: Performance spans
Month 1: Business metrics

Future-Proofing

Use OTel-compatible span/attribute names now (zero cost, easier migration later):

• http.request.method not method
• ui.screen.load not screenLoad
• app.start.cold not coldStart

See references/otel-mobile.md for full naming conventions.

Anti-Patterns

• Measure everything (noise, battery drain)
• Skip symbolication (unreadable crashes)
• Block main thread for telemetry
• Log PII in breadcrumbs

Implementation

See references/instrumentation-patterns.md for:

• Detailed 5-tier checklist
• Span naming conventions
• Sampling strategies
• Testing checklist

See references/jtbd.md for full Jobs-to-be-Done framework.

Related Skills

• See skills/user-journey-tracking for implementing JTBD instrumentation
• See skills/crash-instrumentation for Tier 1 (crashes)
• See skills/symbolication-setup for Tier 1 (readable stack traces)