AI Visibility Tracker: Spreadsheet Template
What this is: the weekly measurement system from Part 5: Measuring Agentic Visibility as a ready-to-use workbook. Five sheets cover all four visibility layers, with the guide's alert thresholds built in as formulas.
How to use it:
- Download the workbook, fill the blue cells on the Setup sheet (product, MCP package, skills repo, your 5 target task queries).
- Add one row to Weekly Log each week. Blue cells are inputs (scriptable); black cells calculate deltas and alerts automatically.
- Log every eval run on Eval Runs; check Dashboard for the latest snapshot and fired alerts.
📥 Download: visibility-tracker.xlsx
What's inside
| Sheet | Contents |
|---|---|
| README | The rules of the road from Part 5: npm is the honest ledger; ±10% on any snapshot; skills.sh counts are order-of-magnitude; quarterly tier re-verification |
| Setup | Your product identifiers + the 5 target task queries (phrased as developer tasks, not brand names) |
| Weekly Log | Positions per query, benchmark score + w/w delta, days-since-update with automatic RE-PARSE alert at >7 days, skills.sh installs, npm weekly downloads + % delta, VS Code gallery check, PulseMCP estimate (secondary trend only), optional mentions column deliberately labeled "mentions," not "AI visibility" |
| Eval Runs | One row per suite run (deprecation / flip-rate / integration / doc-QA): model + version, trigger, result vs target (deprecation target: <5% emission), regression flag, raw-results link |
| Dashboard | Latest values pulled from the log with alert formulas: benchmark −3pts w/w, freshness >7 days, missing VS Code listing, plus the quarterly tier-re-verification reminder |
The thresholds, and where they come from
Every alert in the workbook traces to a finding in the guide, data as of 2026-06-11: the 7-day freshness alert comes from freshness being the strongest quality-score correlate we measured (Spearman ρ=−0.54, n=17); the npm-as-ledger rule comes from the ~99% directory-bypass finding; the <5% deprecation target comes from Play 8; the weekly (not quarterly) cadence comes from watching a top-10 library lose half its retrieval share in 30 days.
Prefer to build your own? The full metric stack and check list is in Part 5. This workbook is just that page with formulas.
Part of The Complete Playbook to Agentic Discovery · from Part 5: Measuring Agentic Visibility.