Purpose
This note defines the actual workbook shape for the pilot evidence tracker.
Workbook name:
AI Enablement - Pilot Evidence Tracker - Initial Pilot Cohort.xlsx
Recommended location:
- restricted SharePoint document library
- or a Teams channel files area with limited membership
Worksheet list
Create these tabs in this order:
Pilot overviewLookup listsWorkflow KPI trackerSampled case logSupport and learning logWeekly action log
Tab 1. Pilot overview
Use one simple metadata block.
Fields:
- pilot name
- pilot start date
- pilot end date
- baseline window
- sponsor
- technical enablement lead
- manager sponsor
- technical reviewer
- survey links
- Jira filter link
- workbook owner
- current pilot decision:
refine,hold,pause,scale
Tab 2. Lookup lists
Use this tab for validation values so the rest of the workbook stays consistent.
Suggested lists:
- roles
- target workflows
- workflow mode
- AI surface
- familiarity values
- risk values
- verification difficulty values
- trend values
- action values
Default lookup values:
Workflow mode
learningdelivery
AI surface
chatIDE assistantrepo-aware assistantother
Familiarity
F1 unfamiliarF2 mixedF3 familiar
Risk
R1 lowR2 mediumR3 high
Verification difficulty
lowmediumhigh
Trend
updownflatmixedunknown
Action
supportinspectrefinepausescalehold
Tab 3. Workflow KPI tracker
Use one row per workflow + KPI + week.
Columns:
| Column | Purpose |
|---|---|
| Week starting | weekly anchor |
| Workflow tag | stable workflow identifier |
| Role | reporting cut |
| KPI | one of the approved KPIs |
| Data source | Jira, PR system, Forms, rubric, other |
| Owner | person updating the row |
| Baseline | numeric or short text baseline |
| Current | current period value |
| Change | current minus baseline when numeric |
| Trend | up/down/flat/mixed/unknown |
| Decision note | what this means |
| Source link | report or export location |
Approved KPI names:
Target workflow median cycle timeDownstream rework rateReview burdenGuardrail exception countAdoption quality scoreRetrieval follow-up scoreConfidence calibration check
Tab 4. Sampled case log
Use one row per reviewed case.
Columns:
| Column | Purpose |
|---|---|
| Week starting | weekly anchor |
| Review date | date reviewed |
| Artifact link | issue, PR, doc, or case link |
| Role | participant role |
| Team | team or cohort group |
| Workflow tag | target workflow |
| Workflow mode | learning or delivery |
| AI surface | tool surface used |
| Task familiarity | F1/F2/F3 |
| Task risk | R1/R2/R3 |
| Verification difficulty | low, medium, or high |
| Workflow fit score | 0-2 |
| Explanation ability score | 0-2 |
| Verification quality score | 0-2 |
| Risk handling score | 0-2 |
| Downstream quality score | 0-2 |
| Total score | sum of the five scores |
| Strongest part | short text |
| Biggest concern | short text |
| Teaching note | short text |
| Follow-up owner | owner if action is needed |
| Escalation needed | yes or no |
Suggested total formula:
=SUM(L2:P2)if the score columns begin inL
Tab 5. Support and learning log
Use one row per support signal or learning event.
Columns:
| Column | Purpose |
|---|---|
| Week starting | weekly anchor |
| Signal type | workshop, office hours, manager check-in, peer support, retrieval check |
| Event or source | session name or source |
| Audience | who it served |
| Count | attendance or response count |
| Top support need | main request or issue |
| Follow-up owner | who handles it |
| Note | short observation |
Tab 6. Weekly action log
Use one row per week.
Columns:
| Column | Purpose |
|---|---|
| Week starting | weekly anchor |
| What improved | short summary |
| What got heavier | short summary |
| What needs support | short summary |
| What needs inspection | short summary |
| Decision | support/refine/pause/scale/hold |
| Owner | who carries the action |
| Due date | target completion |
| Status | open or closed |
Workbook rules
- keep raw survey exports outside this workbook
- keep names out of the shared tabs by default
- use links to internal cases instead of copying sensitive content
- if one tab stops informing decisions, simplify it instead of preserving it for completeness
Minimum useful formulas
Total scorein the sampled case log- optional conditional formatting for low scores and escalation flags
- optional simple pivot or filter view by workflow
Do not over-automate this workbook in the first pilot.
The point is operational clarity, not analytics theater.