Release Plan · v1.0
A sequence of trust-earning gates, not a launch date
The PRD's three-phase scope, operationalized into entry/exit gates. Each phase expands scope only after the last one earns it.
Owner Product, with Engineering & Design · per-account "pause AI" kill-switch exists from day one.
Three phases
Phase 1
MVP — Creative + Suggest
Meta only. 5–10 design partners, $50K+/mo.
Exit
- ≥40% acceptance, sustained 4+ weeks
- Zero brand-safety failures reaching a live account
Phase 2
Multi-platform + budget/bid
+Google Ads. Budget/bid recs, still Suggest-only.
Exit
- Budget/bid acceptance reaches parity with creative
- Positive incremental-ROAS trend across cohort
Phase 3
Auto-apply lane
Opt-in only, per campaign, trust-tier eligible accounts.
Success
- Auto-apply ≥ own Suggest-lane ROAS history
- Rollback rate low and stable
Feature flags
| google_ads | Off until Phase 2 entry |
| budget_bid | Off until Phase 2 |
| auto_apply | Off until Phase 3, then opt-in per account |
| ai_paused | Per-account kill-switch, available from Phase 1 |
Post-launch monitoring
- Weekly review of guardrails: LLM cost %, acceptance rate, escalation rate.
- Confidence calibration is ongoing release-health work, not a one-time launch task.