Methodology
How Aoraforge measures AI search citation, and why every number carries an interval.
The statistics, sample-size discipline, AI-search-engine independence, and verification logic that underpin every report we ship. Read in any order; the essays cross-reference each other.
Essay 01
Wilson 90% confidence intervals, on every measurement
Why every citation rate Aoraforge reports carries a real interval, not a single number — and why we use Wilson rather than the normal-approximation interval that other audits default to.
Essay 02
Four AI platforms, measured independently
Why aggregating ChatGPT, Perplexity, Google AI Overviews, and Brave Search into one citation rate hides the structure of the gap, and what each AI search engine actually rewards.
Essay 03
Four check-ins across 30 days, not a single snapshot
AI search indexes update on different cadences. A single Day-30 measurement cannot separate "the work landed" from "the index hadn't crawled yet." Aoraforge polls four times across the verification window so the trajectory becomes the audit.
Essay 04
Pre-registered queries — committed before any work begins
Why Aoraforge locks the 15 target queries before any work begins, and how the end-of-cycle manual review against the polling logs keeps the standard reproducible.