Measurement plan · Updated September 12, 2026
Measure reader value without collecting more than needed.
No analytics script is active on this build. This page defines the event vocabulary and decision rules for a future consented, privacy-reviewed implementation; it does not collect prompts, model outputs, or workload contents.
What will be measured
| Event | Reader action it represents |
|---|---|
| verdict_view | A reader reaches a published, evidence-backed verdict. |
| evidence_expand | A reader opens the proof, limits, or method behind a conclusion. |
| artifact_download | A reader retrieves a non-sensitive research artifact. |
| related_comparison_click | A reader continues to a relevant decision page. |
| verdict_alert_signup | A reader requests a future verdict-change alert. |
| newsletter_signup | A reader joins a consented Buyer Brief. |
| outbound_source_click | A reader checks a cited source. |
Decision rules
Use aggregate Search Console and Bing data to see discovery, page-level evidence engagement to see usefulness, and returning readers or opt-ins to see retention. Compare meaningful 28-day periods only after enough traffic exists and annotate releases. Search position, AI citation, and a single campaign spike are diagnostics—not proof of reader value.
What must happen before collection
A verified account owner, privacy/consent review, and a compliant destination for email must exist. Until then, RSS remains the available return channel and this site will not present a fake signup form or silently add tracking.
Follow the Atom feed · read editorial standards · see the research queue