TL;DR: Complete pre-launch checklist for any AI implementation: strategy (6 items), data (5), technical setup (6), team readiness (5), testing (3), launch (2). Built from dozens of deployments. Every item exists because we've seen what happens when it's skipped.
---
## Phase 1: Strategy and Scope
1. Define the specific business problem in one sentence. 2. Set 3-5 measurable success criteria (KPIs). 3. Scope the pilot audience — one team, one workflow. 4. Get executive sponsorship — one leader owns this. 5. Set the timeline with milestones. 6. Define budget including 15% contingency.
## Phase 2: Data Preparation
7. Audit every data source the AI will access. 8. Clean data — duplicates, formatting, outdated records. Prevents 50% of quality issues. 9. Fill knowledge gaps — undocumented answers get documented now. 10. Standardize formats across sources. 11. Set up data access and permissions.
## Phase 3: Technical Setup
12. Configure AI model — parameters, system prompts, use case tuning. 13. Build integrations — CRM, email, scheduling, knowledge base. 14. Set up automation workflows — triggers, conditions, actions. 15. Configure error handling — fallback paths for every failure mode. 16. Implement logging — inputs, outputs, response times, error rates. 17. Set up feedback mechanism — make it easy for users to flag bad outputs.
## Phase 4: Team Readiness
18. Build role-specific prompt packs. 19. Create one-page quick-start guide. 20. Schedule hands-on training workshops (not presentations). 21. Brief 1-2 AI champions per team. 22. Communicate launch plan one week before go-live.
## Phase 5: Testing
23. End-to-end testing with real data. Every trigger, condition, output. 24. Edge case testing — unusual inputs, empty fields, out-of-scope questions. 25. Pilot deployment — monitor 48 hours, daily feedback.
## Phase 6: Launch
26. Go live with monitoring dashboard active. Someone watching metrics for first 24 hours. 27. Schedule 30-day review — metrics, feedback, optimization plan.
## Quick Reference
Strategy: Problem. KPIs. Pilot. Sponsor. Timeline. Budget. Data: Sources. Clean. Gaps. Formats. Access. Technical: Model. Integrations. Workflows. Errors. Logging. Feedback. Team: Prompts. Guide. Training. Champions. Comms. Testing: End-to-end. Edge cases. Pilot. Launch: Monitoring. 30-day review.
## FAQ
Q: Need all 27? A: Simple automation — maybe skip a few. Anything customer-facing or multi-team — all 27. Skipped items bite you.
Q: How long to complete? A: 2-4 weeks. Data prep takes longest.
Q: Most commonly skipped? A: Error handling (#15) and feedback mechanisms (#17). Both essential.
Q: AI champion — manager? A: Ideally not. Enthusiastic IC respected by peers is more effective.
Q: Fail during testing? A: Fix and retest. Never go live with known failures.
Q: Run items in parallel? A: Strategy and data overlap. Technical starts at 80% data prep. Team readiness starts when technical is confirmed. Testing is last, sequential.
---
Want help with this checklist? Strategy Audit covers items 1-6 and plans the rest. Our team handles 7-27. [Book your Strategy Audit →](/get-started)