Stage 9: Feedback Loop
Stage 9: Feedback Loop
Owner: Fiona (Customer Feedback Analyst) Blueprint stage: Stage 9 -- Feedback Loop
Trigger
Post-delivery, ongoing during support period and beyond.
Inputs
| Source | What |
|---|---|
| Dana (Stage 7) | Initial client feedback from handover |
| Client | Ongoing usage feedback |
| Support log (Stage 8) | Support requests and resolutions |
Activities
9.0 Feedback Channel Setup
The feedback channel is established by Dana during Stage 7 (Delivery). Fiona activates monitoring at the start of Stage 9.
| Product Type | Channel | Mechanism |
|---|---|---|
| Agent (repo-based) | feedback/ folder in deployed repo |
Customer creates one markdown file per issue using feedback-template.md. Fiona monitors the folder. |
| Standalone (HTML tool, document) | Shared delivery folder | OneDrive/agents/{client}/feedback/ or equivalent. Customer drops feedback files there. |
| Any product | Email fallback | Dedicated support email (if configured in Stage 8). Dana forwards to Fiona. |
Template: workflows/templates/feedback-template.md — covers bug reports, feature requests, questions, and usability issues with priority levels.
Monitoring cadence: Fiona checks active feedback channels weekly during the support period, or on notification from Dana/support.
9.1 Define Feedback Schedule
| Milestone | When | Method |
|---|---|---|
| Initial | At delivery (Stage 7) | Collected by Dana |
| 30-day | 30 days post-delivery | Structured questionnaire |
| 60-day | 60 days post-delivery | Structured questionnaire |
| 90-day | 90 days post-delivery | Questionnaire + review call |
| Ongoing | As needed | Ad-hoc via support channel |
9.1b Autonomy Promotion Review (Bike Method)
The 30/60/90-day milestones double as autonomy promotion reviews. See departments/methodology/methodology-specialist-marcel/manufacturing-methodology/autonomy-ladder.md.
At each milestone, Fiona checks the talent's performance against the promotion criteria for its current rung:
- Rung 1 → 2: ≥10 supervised runs, ≥90% outputs accepted without substantive correction, 0 critical errors.
- Rung 2 → 3: clean window, sustained acceptance, 0 critical regressions, client sign-off.
Outcome is promote / hold / demote. A critical error drops the talent one rung immediately (posture in settings.json tightened to match). Pablo approves Rung 1→2; the client approves Rung 2→3. The new rung, the evidence, and the settings.json change are recorded in the order's process/.
9.2 Feedback Questions
At each milestone, ask about each capability delivered:
- Frequency of use (daily, weekly, rarely, never)
- Output quality (1-5)
- Missing features or gaps
- Usability issues
- Overall satisfaction (1-10)
- Top change request
- NPS recommendation score
Product-specific feedback questions come from the work order.
9.3 Analyze and Route Feedback
| Feedback Type | Route To | Action |
|---|---|---|
| Product defect | Max (Maintenance) | Bug fix |
| Feature request | Pablo (Production Line) | Evaluate for next version |
| Domain/methodology gap | Pablo + Ada | Pattern/skill enhancement |
| Usability issue | Pablo | Skill prompt improvement |
| Documentation gap | Remy (Docs) | Documentation update |
| Process issue | Relevant department | Process improvement |
| Positive feedback | Company records | Testimonials and learning |
9.4 Feed Back into Factory
Aggregate insights across clients to improve: the production line, pattern catalog (Ada), quality gates (Quinn), assembly process (Pablo), and intake questions (Camille).
Outputs
| Deliverable | Format | Destination |
|---|---|---|
| Feedback reports | Markdown | Order records |
| Improvement recommendations | Markdown | Routed to relevant roles |
| Client satisfaction score | Numeric | Company dashboard |
Quality Gate
Gate: Feedback collected and actionable items routed. Criteria: Schedule defined, first feedback collected, actionable items routed, satisfaction baseline established. Evidence: Feedback log + routing log.