Skip to content

Add product analytics as an evidence source and experiment-design mechanism#2

Open
bfish1996 wants to merge 1 commit into
mainfrom
claude/telemetry-data-integration-o5uqgh
Open

Add product analytics as an evidence source and experiment-design mechanism#2
bfish1996 wants to merge 1 commit into
mainfrom
claude/telemetry-data-integration-o5uqgh

Conversation

@bfish1996

Copy link
Copy Markdown
Owner

Teams using validation-os often already have PostHog/Mixpanel/BigQuery
telemetry, but their event schemas are workspace-defined and unknowable in
advance, so this composes with each platform's own official agent skill
(PostHog/ai-plugin, mixpanel/ai-plugins, or an equivalent from skills.sh)
for schema discovery and querying rather than reimplementing it.

  • analytics-metric-resolution.md: shared glossary-check → introspect →
    gated-disambiguation procedure, reused by both directions
  • quant-analytics-rubric.md: the quantitative-data trust rubric (schema
    drift, denominators, segment size, confounds, no post-hoc cherry-picking)
  • find-evidence/references/product-analytics.md: pulls a measured metric
    in as revealed-tier evidence
  • experiment-design/references/analytics-instrumentation.md: pre-registers
    a live query/metric as the pass/kill bar, optionally wiring a platform's
    native A/B or feature-flag tools
  • config/setup/docs updated to declare the new analytics evidence source
    and two new worked examples (08, 09) on the existing Ledgerly thread

…hanism

Teams using validation-os often already have PostHog/Mixpanel/BigQuery
telemetry, but their event schemas are workspace-defined and unknowable in
advance, so this composes with each platform's own official agent skill
(PostHog/ai-plugin, mixpanel/ai-plugins, or an equivalent from skills.sh)
for schema discovery and querying rather than reimplementing it.

- analytics-metric-resolution.md: shared glossary-check → introspect →
  gated-disambiguation procedure, reused by both directions
- quant-analytics-rubric.md: the quantitative-data trust rubric (schema
  drift, denominators, segment size, confounds, no post-hoc cherry-picking)
- find-evidence/references/product-analytics.md: pulls a measured metric
  in as revealed-tier evidence
- experiment-design/references/analytics-instrumentation.md: pre-registers
  a live query/metric as the pass/kill bar, optionally wiring a platform's
  native A/B or feature-flag tools
- config/setup/docs updated to declare the new `analytics` evidence source
  and two new worked examples (08, 09) on the existing Ledgerly thread
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants