CM-201b · Module 1

Designing for Replication

4 min read

A pilot that works but cannot be replicated is a demonstration, not a proof. The most common version of this failure: the pilot succeeds because of a specific champion, a specific team culture, and a specific workflow context. Those three variables are not transferable. When the rollout starts, the champion is busy, the team culture is different, and the workflow does not quite match. The results do not replicate. The skeptics say they knew it would not work at scale.

Designing for replication means treating the pilot documentation as the primary deliverable, not the results dashboard.

  1. 1. Document the Workflow Before and After Capture exactly how the work is done before AI: steps, time, tools, handoffs, error rates. Then capture exactly how it is done with AI. The delta is the value story. The after-state description is the replication target.
  2. 2. Document the Training Approach What training did the pilot team receive? In what format, over what time period, with what level of ongoing support? This is the training playbook for the rollout. If you cannot reproduce the training, you cannot reproduce the results.
  3. 3. Document the Champion Profile What made the pilot champion effective? Their credibility, their communication style, their relationship to the team, their specific advocacy behaviors. Use this profile to identify champion candidates for the rollout cohorts.
  4. 4. Document What Failed and How It Was Resolved Every pilot has problems. Document them and the resolutions. A rollout team that encounters the same problem and has a documented solution is a rollout team that recovers. A rollout team that encounters the same problem without documentation is a rollout team that escalates to the Executive Sponsor.
  5. 5. Validate Replicability Before Declaring Pilot Complete Before closing the pilot, have a second team attempt to replicate the results using only the documentation — no access to the original champion, no special setup. If the second team succeeds, the playbook is complete. If they struggle, the documentation needs refinement.