Signal Registry
This is the canonical list of metrics we publish every cycle. Log a new signal or Run the Cloudflare analytics export. Exit metric: registry stays at ≤8 signals and every entry refreshes within the 30-day SLA.
Signals Table
All active signals share the same SLA: refresh every 30 days or raise an exception. The cap stays at eight.
| id | Name | Source | Owner (@handle) | Refresh after days | Direction | Threshold / stop rule | Kill criteria |
|---|---|---|---|---|---|---|---|
m-nav-open | Navigation open rate | Cloudflare analytics export (nav_open event via scripts/cloudflare-analytics.mjs) | @lop | 30 | Up good | >= baseline + show 10% lift in pilot tags | Remove if nav instrumentation drops for >1 cycle |
m-time-to-answer | Time to answer | Cloudflare analytics export (time-to-answer event) | @lop | 30 | Down good | <= 60 seconds for 80% of tracked sessions | Drop if CTA instrumentation is disabled |
m-lab-pass | Annex lab pass rate | reports/labs.json (Quick-Run + Proof Run) | @lop | 30 | Up good | >= 0.9, freeze if < 0.8 | Remove when Verify-in-10 labs retire |
m-defect-rate-changed-pages | Defects per changed page | Link Drift + QA follow-ups recorded in reports/labs.json | @lop | 30 | Down good | <= 0.05; investigate if > 0.08 | Drop if defect tracking moves to another source |
m-dashboard-freshness-days | Dashboard freshness | Manual value in reports/cloudflare-export.json > dashboard_freshness | @lop | 30 | Down good | <= 30 days; exception at 35+ | Remove if dashboards merge with State |
m-decision-hit-rate | Decision hit rate | Governance receipts + decision ledger | @lop | 30 | Up good | >= 60% by second tag | Retire if decision receipts move to tooling |
m-time-to-freeze | Time to freeze | Exceptions log + freeze manager | @lop | 30 | Down good | <= 24 hours to activate a freeze trigger | Remove if mitigations move to another platform |
m-time-to-recovery | Time to recovery | Exceptions log + Quick-Run receipts | @lop | 30 | Down good | <= 7 days to exit freeze | Remove if recovery tracking automates elsewhere |
How to add a signal
- Gather the source and ensure it meets the privacy/ROI guardrails.
- Add a new row to the table above with:
- id: machine-readable key.
- Name: human label.
- Source: document the export or script.
- Owner: GitHub handle of the steward.
- Refresh after days: SLA for updating it.
- Direction: “Up good” or “Down good.”
- Threshold: When to raise a flag.
- Kill criteria: When to stop collecting the signal.
- Update any automation (scripts, workflows) that emits the signal.
- Reference the new signal from Receipts or the relevant decision entry.