Eve library

OSHUN Post-Launch Hardening (First 30 Days)

Spec

Revision 1

On this page

Week 1: Stabilize and Observe#

Metrics Review Cadence#

Daily review of key metrics at the 09:30 standup:

Metric Target Source
Crash rate (iOS) < 0.5% Sentry
Crash rate (Android) < 0.5% Sentry + Play Console Vitals
ANR rate (Android) < 0.2% Play Console Vitals
BFF error rate < 1% BFF Prometheus metrics
BFF P95 latency < 500ms BFF Prometheus metrics
Onboarding completion >= 85% Analytics: onboarding.completed
D1 retention >= 40% Analytics: session.start D+1
Domain launch rate >= 70% Analytics: domain.launched
Search usage >= 20% Analytics: search.executed

Activation Tracking#

A user is "activated" when they complete all of:

  1. Account creation.
  2. Onboarding flow completion.
  3. At least one domain launch.
  4. Return visit on D+1.

Track activation rate as the primary leading indicator of product-market fit.

Alert Tuning#

After 7 days of production data:

  • Review alert noise-to-signal ratio. Reduce false positives.
  • Adjust thresholds based on observed baselines.
  • Add new alerts for patterns discovered during launch (e.g., specific API endpoints with elevated error rates).

Week 2: Identify and Fix UX Friction#

Friction Identification Methods#

  1. Funnel Analysis: Identify drop-off points in:

    • Onboarding flow (step-by-step completion rates)
    • Home → domain launch conversion
    • Search → result tap conversion
    • Profile → settings engagement
  2. Session Recordings (if consented): Review anonymized session replays for confusion patterns, dead-end navigations, and rage taps.

  3. Support Ticket Analysis: Categorize incoming support requests:

    • Navigation confusion
    • Feature discoverability
    • Performance complaints
    • Content quality feedback
  4. Beta Survey Follow-Up: Re-survey beta cohort members who gave low NPS.

  5. App Store Reviews: Monitor and respond to 1-3 star reviews. Extract actionable themes.

Top 10 UX Friction Fix Process#

  1. Collect all friction signals from the methods above.
  2. Score each by: frequency × severity × fix effort.
  3. Rank and select top 10.
  4. Assign owners and target ship dates.
  5. Ship fixes via OTA updates (JS-only) or expedited store releases.
  6. Verify improvement via before/after metrics.

Template for tracking:

# Friction Point Frequency Severity Effort Owner Status Ship Date
1 [Description] High/Med/Low P1-P3 S/M/L [Name] [Status] [Date]

Week 3: Experiment and Optimize#

Recommendation Quality Tuning#

The home feed recommendation engine needs calibration based on real usage data:

  1. Baseline measurement: Record current metrics:

    • Home feed tap-through rate (CTR)
    • Domain balance (are all 4 domains getting engagement?)
    • Scroll depth (how far do users scroll before stopping?)
    • Time-of-day patterns (do evening users engage more with Nyx?)
  2. Signal weighting adjustment:

    • Increase weight for domains the user has explicitly favorited.
    • Decrease weight for domains the user consistently ignores.
    • Add recency bias: prioritize fresh content over stale.
    • Add time-of-day signals: Tara mornings, Nyx evenings, Arete check-in times.
  3. A/B test iterations:

    • Experiment: home-feed-ranking-v2 — test new signal weights.
    • Traffic: 10% for 7 days.
    • Primary metric: home feed CTR.
    • Guardrails: session duration drop, domain diversity index.

Notification Quality Tuning#

  1. Opt-in rate tracking: What percentage of users enable notifications?
  2. Tap-through rate: Which notification types drive re-engagement?
  3. Dismiss rate: Which notifications are ignored or dismissed?
  4. Unsubscribe rate: Are notifications causing users to disable them?

Tuning actions:

  • Reduce notification frequency for users with high dismiss rate.
  • Personalize notification timing based on user's active hours.
  • A/B test notification copy for Arete daily check-in reminders.
  • Remove low-performing notification types.

Week 4: Review and Roadmap Rebalance#

30-Day Metrics Report#

Compile a comprehensive report covering:

  1. Acquisition:

    • Total installs (iOS + Android + web)
    • Install-to-signup conversion
    • Signup-to-activation conversion
    • Acquisition channel breakdown
  2. Activation:

    • Onboarding completion rate
    • First domain launch rate
    • Activation rate (defined above)
    • Time-to-activation (median)
  3. Retention:

    • D1, D7, D14, D30 retention curves
    • Retention by domain preference
    • Retention by onboarding path
    • Cohort comparison (Week 1 vs. Week 2 vs. Week 3 signups)
  4. Engagement:

    • DAU/MAU ratio
    • Sessions per DAU
    • Domain diversity (avg domains used per user)
    • Cross-domain journey frequency
    • Feature adoption rates (search, favorites, offline, sharing)
  5. Quality:

    • Crash-free session rate
    • App Store / Play Store rating
    • Support ticket volume and resolution time
    • NPS score (from in-app survey)
  6. Performance:

    • Cold start P50/P95 (iOS and Android)
    • Home feed TTI P50/P95
    • Domain launch P50/P95
    • BFF latency P50/P95/P99

Roadmap Rebalance#

Based on the 30-day report:

  1. Kill underperforming features: If a feature has < 5% adoption after 30 days and isn't trending up, consider removing it or redesigning.

  2. Double down on winners: Features with high engagement and positive feedback should get investment priority.

  3. Address retention gaps: If D7 retention is below target, prioritize the friction fixes that most impact the Day 2-7 experience.

  4. Domain balance: If one domain has significantly lower engagement:

    • Investigate: is it a discovery problem or a content quality problem?
    • Adjust home feed prominence to increase exposure.
    • Partner with domain team to improve domain-specific value proposition.
  5. Update POST_MVP_SCOPE.md: Re-prioritize Wave 1-3 features based on real usage data. Move high-demand features up. Deprioritize features that user data doesn't support.


Domain-Specific Acceptance Verification#

Tara#

Criterion Method Target
Launch meditation from home in <= 2 taps Beta script 2 100% pass
Resume session syncs across devices Cross-device test No data loss
Audio playback survives interruptions Manual test Graceful resume
Audio plays in background Manual test Continues
Favorites persist after reinstall Reinstall test All preserved

Veritas#

Criterion Method Target
Open top story in <= 2 taps Beta script 2 100% pass
Confidence indicators visible on small screens Device matrix test Readable SE
Save/follow persists across mobile and web Cross-platform test Synced
Evidence trail is navigable Manual test Tap-accessible

Nyx#

Criterion Method Target
Access nightly highlights quickly Beta script 3 < 3 seconds
Core content usable in low-connectivity Offline test Cached content
Night-mode ergonomics are comfortable User survey >= 80% positive
Sky map renders without jank Performance test < 16ms frames

Arete#

Criterion Method Target
Daily check-in in <= 60 seconds Beta script 3 100% pass
Goals/habits/journal integrated in shell Manual test All accessible
Progress and streak visuals are clear User survey >= 80% positive
Journal entries persist and sync Cross-device test No data loss

Cross-Platform Definition of Done#

Platform Check Method Status
iOS Smoke tests pass Maestro e2e [ ]
iOS Accessibility audit passes axe / VoiceOver [ ]
iOS Performance budgets met Startup + jank [ ]
Android Smoke tests pass Maestro e2e [ ]
Android Accessibility audit passes axe / TalkBack [ ]
Android Performance budgets met Startup + jank [ ]
Web Responsive layout verified Playwright viewport [ ]
Web Accessibility audit passes Lighthouse >= 90% [ ]
Web SEO checks pass Lighthouse >= 85% [ ]
PWA Installability verified manifest.json [ ]
PWA Offline mode works sw.js test [ ]
PWA Update flow works SW update message [ ]
All Shared auth verified Cross-surface test [ ]
All Entitlements synced Subscription test [ ]
All Analytics events validated Event audit [ ]
All Monitoring and alerts live Dashboard check [ ]
All Runbooks documented ON_CALL_RUNBOOK.md [ ]
All Documentation up to date docs/releases/v1/* review [ ]