EST. MMXXVI · THE INSTITUTION OF RECORD FOR AI CITATION · OPEN METHODOLOGY
the blog
Findings from our own runs — not folklore. Every number here comes from a panel we can point you at, with the caveats attached.
On 2026-08-08 we ran the same six questions through ChatGPT, Perplexity, Gemini, and Google — the surfaces real buyers use, not our measurement API. Google's AI Overview praised our published honesty verbatim eight days after launch. Consumer Gemini could not find us and grouped us with 'link networks or simple prompt injection.' Here are all the receipts, including the ones that hurt.
read the finding →For fifteen weeks we have measured one business — our own auction-software product — on the same AI-visibility instrument we sell. Here is the entire run ledger: every completed scan, the two that failed, the week a parser bug inflated a number, the platform shifts that moved every brand at once, and how we handle a panel that keeps growing. Nothing retouched.
read the finding →On 2026-08-08 the Internet Archive held zero snapshots of listerleo.com for the domain's entire lifetime. We asked it to save five URLs; each capture recorded a 404, while the same URLs served a normal 200 page to every user agent we tested. Here is the trace, the method, and what would overturn it.
read the finding →AI answers about AI-visibility tools are built from about 19 comparison articles. We audited the ten most-cited: roughly one in four of their statistics carries any source, and about 3% of third-party listicle claims trace to a primary source. Here is the data, our own pages included.
read the finding →Across nine businesses on our frozen five-engine panel, AI assistants named them in nearly 100% of answers when asked by name, and volunteered them in as few as 0% of category answers. The gap between those two numbers is the whole game.
read the finding →The AI Citation Institute measures whether AI assistants recommend a business — and on day one, on our own panel, we appeared in 0.0% of the answers. Here's what we measure, and why we published our own zero.
read the finding →We audited all 91 of our own blog posts across four AI engines. The pages nobody linked to internally were cited almost never. The pages with ten or more inbound links were cited about 44% of the time.
read the finding →