Category: Uncategorized

Tuesday we pre-registered four predictions about which reputation signal separates an AI “consensus” pick from an equally-good runner-up: best-of breadth, review rating/volume, and a breadth-gated moat. We scored 18 brands across 6 buyer-intent categories against those predictions — and three of the four failed. Breadth doesn’t separate the pick (runner-ups sat on as many or…

Read more →

Monday we proved the AI “consensus” pick is real (two engines’ shortlists overlapped in all 12 buyer-intent categories) but can’t be self-manufactured (self-citing vendors won 0 of 2). So which reputation signals actually separate a consensus pick from an equally-good runner-up? Here’s our full pre-registered method: we score three candidate signals — third-party best-of breadth,…

Read more →

The 2026 advice is that you become the AI “consensus” pick by piling up third-party mentions and reviews — 85% of brand mentions come from off-site sources, and review-rich brands get recommended more. Our first-party data confirms a consensus even exists: across 12 buyer-intent categories, two engines’ top recommendation sets overlapped 100% of the time…

Read more →

The advice says publish your own “best-of” listicle, rank yourself #1, and win the AI recommendation. Our week-long GEO Lab experiment says the self-ranking trap is real — and worse than advertised. A big vendor’s own listicle was cited in only 8.3% of 24 buyer-intent answers, the #1 recommended brand’s own site was the cited…

Read more →

We scored 24 buyer-intent answers from Gemini and Perplexity to test whether publishing your own “best-of” listicle and ranking yourself #1 wins the AI recommendation. It doesn’t — and it barely wins a citation. A major vendor’s own listicle was cited in just 8.3% of answers, the #1 recommended brand’s own site was the cited…

Read more →

Monday we showed self-published “best-of” listicles get cited by AI engines — but a 100-query study found the recommendation goes to a listed competitor 69% of the time. So does self-ranking actually pay off? Here’s our full, pre-registered method to separate a citation from a recommendation: a self-serving-listicle audit, a self-published-vs-third-party head-to-head, and a self-ranking-position…

Read more →

The advice is everywhere: publish a “Top 10” listicle with yourself at #1 to get cited by AI. Our first-party data confirms self-published listicles do get cited — vendors earn 59.5% of citations, and 62% of those are their own best-of pages. But a new 100-query study shows those citations recommend a competitor 69% of…

Read more →

We dissected 306 AI citations across 219 domains to find what the domains engines choose have in common. The surprise: by domain TYPE, vendor own-sites dominate (59.5%) — but by page FORMAT, 66% of every citation is a “best-of” listicle or comparison, and even vendors earn 62% of their citations from their own roundup content,…

Read more →

Monday we showed 15 domains don’t control AI citations for buyer-intent queries — the citation basket is a 220-domain long tail. So what do the domains engines DO reach for have in common? Here’s our full, pre-registered method: a domain-type taxonomy, a page-format audit, and an authority-vs-structure test on 220 first-party cited domains. Four predictions…

Read more →

A new industry index of 680M citations claims 15 domains capture 68% of all AI citation share — “more extreme than PageRank.” We reran the concentration math on our own first-party data for buyer-intent queries. Top 15 = 23%, 220 distinct domains, 77% cited only once. The concentration story is real for general knowledge, not…

Read more →