The AI Visibility Factors, Edition One. Measured, Not Surveyed.
Published 2026-08-03 by Namebeam. Built from 212 classified citations across 43 dated check transcripts plus one clean API run (2026-07-31 to 2026-08-02). Sample sizes printed. Updated on the First Friday cadence as the corpus grows. The industry's existing factor reports are expert opinion surveys; every number below was measured from what AI engines actually cited, and our own losing scores are part of the dataset.
Factor weights for LOCAL BUSINESS queries (what gets a business named)
From 99 citations across 5 real business checks:
- The business's own website: 49.5 percent of citations. The dominant factor by a wide margin. Plain, answer-shaped pages.
- Directory and profile listings: 20.2 percent. The free ring: BBB (cited 4 times for our winners), Yelp (3), city government and utility pages (6), Nextdoor, Mapquest.
- Community and social: 15.2 percent. Facebook pages (3 citations for winners), local threads.
- Roundup listicles: 2.0 percent. Nearly irrelevant for local recommendations, despite dominating marketing conversation.
- News 1.0 percent; unknown and other 12.2 percent.
Both businesses named by all 4 engines carried the own-site-plus-ring pattern. All three zero-score businesses lacked it.
Factor weights for CATEGORY AND TOOL queries (how buyers find vendors like us)
From 93 citations on our own category checks: vendor websites 40.9 percent, community 24.7 percent, roundup listicles 17.2 percent, news 7.5 percent. A different problem than local visibility, which is why one-size fix lists fail.
The personalization factor almost nobody discloses
Same buying query, same day, two channels: signed-in consumer app, roundups were 30.6 percent of citations; clean API channel with personalization stripped, roundups were 17 of 20 citations, 85 percent. Signed-in checks flatter you. Any vendor showing perfect scores without disclosing channel and account state is showing you a mirror, not a measurement.
Convergence with expert opinion
The strongest opinion-based work in the adjacent field points the same direction: Whitespark's Darren Shaw, steward of the long-running Local Search Ranking Factors survey, wrote in Search Engine Land (2026-05-26) that AI search visibility runs through reviews, citations, and third-party mentions. Our transcripts measure exactly that ring. Respect to the survey tradition; we publish the measured layer beneath it.
Method and limits, printed
Engines checked: ChatGPT, Gemini, Perplexity, Google AI Overview, and Claude (1 completed run, the newest addition). Consumer-app runs are flagged when signed in; one row was voided outright for memory contamination and stayed voided. Sample: 5 businesses, 1 self-check, 212 classified citations. Small and growing; every new check updates these weights, wins and zeros alike. Raw corpus: public on this site and on GitHub.