ux: investigate D1 retention drop — 21% to 9.4% after 2026-02-02 growth burst

[agents-squads]

Context

Weekly user success report (2026-04-01) shows D1 retention dropped from 21% (baseline, analytics-14) to 9.4% — a -11.6pp decline, despite 68% growth in total users (19 → 32).

Signal

• Week of 2026-02-02: 16 new users joined (likely from a launch/post)
• Most of these users had only 1-3 runs total
• Only 3 of 32 users returned the day after first use

Hypothesis

1. Onboarding friction: New users hit a wall after first run and don't return
2. Audience mismatch: The Feb 2 burst brought casual explorers, not target users
3. Version regression: CLI change around that date may have broken the getting-started path

Ask

Audit the onboarding flow for users on their first run:

• What does the output look like for a brand-new user with no config?
• Are there clear next steps shown after squads run?
• Is there a getting-started command or guide surfaced on first run?

File sub-issues for any friction points found.

Report: customer/analytics/user-success-2026-04-01.md
Related: analytics#14 (baseline), squads-cli #661 (Docker Compose gap — another onboarding friction point)