This is an open skill pack for AI agents to read web analytics. Covers MECE hypothesis trees, sample-size math, Simpson's paradox, funnel drop-off ranges, and metric benchmarks by business model.
Seven skills:
- analytics-profile-setup: 5-minute interview that calibrates to your business
- analytics-diagnostic-method: the spine, MECE + Pyramid Principle
- traffic-change-diagnosis: hypothesis tree for "why did traffic drop"
Author here. I work on Clamp Analytics.
This is an open skill pack for AI agents to read web analytics. Covers MECE hypothesis trees, sample-size math, Simpson's paradox, funnel drop-off ranges, and metric benchmarks by business model.
Seven skills:
- analytics-profile-setup: 5-minute interview that calibrates to your business
- analytics-diagnostic-method: the spine, MECE + Pyramid Principle
- traffic-change-diagnosis: hypothesis tree for "why did traffic drop"
- channel-and-funnel-quality: vanity-traffic detection, expected drop-offs
- metric-context-and-benchmarks: what's a good X, by model
- event-schema-author: declares typed events from existing track() calls
- experiment-result-reader: reads A/B tests with sample-size discipline
Works with Amplitude, Clamp Analytics, GA4, Mixpanel, and PostHog.
Built on the open Event Schema spec (clamp-sh/event-schema).
Happy to answer anything.