AI 扩展计划 / Playbooks
ABPA 模板 10：Adoption Dashboard
95 行abpa/templates/10-adoption-dashboard.md
Adoption Dashboard
Use this to prove that an AI product is becoming part of real work. Adoption is not just logins. Track usage, trust, quality, and behavior change.
Dashboard Context
| Field | Answer |
|---|---|
| Product / capability | |
| Target users | |
| Workflow step | |
| Pilot start date | |
| Adoption owner | |
| Reporting cadence |
Adoption Funnel
| Stage | Definition | Metric | Baseline | Target | Current | Owner |
|---|---|---|---|---|---|---|
| Eligible users | users who should use the tool | count | ||||
| Activated users | used it at least once in real workflow | % eligible | ||||
| Repeat users | used it in 2+ periods | % activated | ||||
| Habit users | use it for most eligible work | % eligible | ||||
| Value users | show measurable quality or efficiency lift | % active |
Usage Metrics
| Metric | Definition | Segment | Target | Current | Interpretation |
|---|---|---|---|---|---|
| Weekly active users | role / team / region | ||||
| Eligible cases touched | case type | ||||
| Suggestions generated | task type | ||||
| Suggestions accepted | task type | ||||
| Suggestions edited | task type | ||||
| Suggestions rejected | task type |
Trust And Quality Metrics
| Metric | Why it matters | Target | Current | Action if bad |
|---|---|---|---|---|
| Acceptance rate | proxy for usefulness | inspect failure cases | ||
| Edit distance / rewrite rate | quality of draft | improve prompt or evidence | ||
| Override rate | trust and control signal | review explanations | ||
| Unsupported claim rate | safety signal | block release and fix grounding | ||
| Escalation rate | workflow fit signal | improve routing or scope | ||
| User confidence score | subjective trust | conduct interviews |
Business Outcome Metrics
| Outcome | Baseline | Target | Current | Evidence source |
|---|---|---|---|---|
| Cycle time per case | ||||
| Cost per case | ||||
| Rework rate | ||||
| Error / defect rate | ||||
| SLA breach rate | ||||
| Quality review score |
Behavior Change Signals
| Signal | Healthy pattern | Unhealthy pattern | Response |
|---|---|---|---|
| Users check cited evidence | cite review becomes normal | blind copy-paste | add UI friction or training |
| Users add feedback | failure cases improve | feedback ignored | close feedback loop |
| Managers discuss metrics | adoption tied to operations | tool seen as side project | steering review |
| Exceptions are explicit | clear escalation | hidden workarounds | workflow redesign |
User Feedback Loop
| Feedback channel | Owner | Cadence | Sample size | Decision it informs |
|---|---|---|---|---|
| In-product thumbs up/down | quality triage | |||
| Structured user interviews | workflow changes | |||
| Support tickets | defect backlog | |||
| Shadowing sessions | process redesign | |||
| Manager review | scaling decision |
Adoption Risks
| Risk | Leading indicator | Mitigation | Owner |
|---|---|---|---|
| Low trust | high rejection or rewrite rate | show evidence and improve evals | |
| Poor workflow fit | users bypass the tool | redesign trigger point | |
| Manager resistance | adoption not discussed in team cadence | align incentives and reporting | |
| Quality regression | more rework or escalations | release gate and rollback | |
| Hidden compliance concern | users avoid sensitive cases | risk review and policy clarity |
Dashboard Review Notes
| Date | Finding | Decision | Owner | Follow-up |
|---|---|---|---|---|