What is the AI Presence report?

A free, manually produced report that measures whether AI assistants actually name your business, and if not, which gate you fall out at. It reads live answers from ChatGPT, Perplexity, Gemini, Claude, and Google AI Mode across several topics and repeated runs, then scores rankings, recognition, references, and reputation to explain the result.

How is it different from the AIO Score?

The AIO Score measures readiness on your own pages: can AI fetch and quote you. The AI Presence report measures the four off-site gates that decide whether AI recommends you, and it reports the observed outcome, your AI Mention Rate. A site can score 97 on readiness and be named in zero answers.

What does it cost and how long does it take?

It is free for now. Because it runs live queries across five engines with repeated runs and a manual review, it takes a few days rather than fifteen seconds. We limit the number we run each week and prioritize businesses with a real service, a real location, and a website we can scan.

A one-paragraph verdict, your AI Mention Rate per engine and per topic, the seven-gate scorecard with the first gate you fail, a developer worklist and a marketing worklist ordered by gate, and an evidence appendix with the captured answers, the queries used, the citation map, and the review matrix.

AI Presence report

Does AI actually name your business?

The AI Presence report answers that question from live AI answers, not from your pages. It reports your AI Mention Rate across ChatGPT, Perplexity, Gemini, Claude, and Google AI Mode, then diagnoses the first gate you fall out at and gives you the fix list in order. It is free for now, produced by hand, and takes a few days.

Why isn’t the AIO Score enough?

Because a perfect readiness score with zero presence is a hygiene score, not a visibility score. We have measured sites that score 97 out of 100 on the free scan and are never named when you ask an AI engine for their category in their city.

The evidence on why is consistent across everything we have read and tested: AI answers are built mostly from off-site sources, models behave like consensus machines, and when sources disagree they play safe and name someone else. The AIO Scorecovers the two gates on your own site. This report covers the five that live off it.

What are the seven gates?

Seven stages of the pipeline every AI engine uses to produce a recommendation, each scored as a gate you either clear or fall out at.

The pipeline is roughly the same everywhere: the engine rewrites the question into sub-queries, retrieves from a search index, judges titles and snippets, reads passages, arbitrates across sources, runs a reputation check, and synthesizes. The diagnosis is the first gate you fail, which is also the order to fix things in.

Gate	Question	What we measure	Where
1 Reach	Can AI fetch you?	robots.txt, firewall, HTTPS, speed, JavaScript dependence	AIO Score (free scan)
2 Readability	Can AI lift a passage from you?	answer up front, question headings, front-loaded sections, static FAQ, pricing stated	AIO Score (free scan)
3 Rankings	Can AI find you in the indexes it grounds on?	top-20 presence in Google and Bing for the sub-queries AI generates for your category	AI Presence report
4 Recognition	Does the machine know who you are?	entity home, one canonical description, profile set, facts that agree everywhere, Knowledge Graph presence	AI Presence report
5 References	Do the sources AI already cites mention you?	presence on the exact pages and domains cited in live answers for your category and city	AI Presence report
6 Reputation	Does AI’s second check clear you?	reviews on three or more platforms, outcome-specific review text, response rate, the “brand + reviews” results, negative threads	AI Presence report
7 Recommendation	Does AI name you?	topic-level mention rate across engines, position in the list, accuracy, persistence over repeated runs	AI Presence report (the headline)

What are the three numbers?

The AI Mention Rate, the AIO Score, and the AI Presence Score, because one number would have to lie.

The AI Mention Rate is the observed outcome: the share of live answers in which you are named, per engine and per topic, from repeated runs, for example “named in 3 of 24 answers, ChatGPT 0 of 6, Perplexity 2 of 6.” The AIO Score is the deterministic readiness score you already know, gates 1 and 2, unchanged. The AI Presence Scoreis new: gates 3 to 6, weighted 25, 20, 35, and 20 points, with off-site references carrying the most because they are the most repeated lever in the evidence.

Those weights are hypotheses. We record them next to observed Mention Rate for every site we run and refit them to whatever actually predicts being named. That calibration will be published here, the same way the AIO Score publishes its research.

How is the report produced?

By hand, from live queries across five engines, repeated, with a person reviewing every answer before anything is written up.

We pick four to six topics for your business (your category in your city, best-of, problem-led, cost, brand due diligence, and comparison), ask each several ways on each engine over repeated runs, and record whether you are named, where in the list, whether you are cited or only mentioned, what was said, and which sources were used. Those sources become your citation map: the exact pages AI uses to answer your category, checked one by one for whether you appear on them.

Rankings are checked for the sub-queries AI generates, in Google and Bing. Recognition is an entity audit across your profiles. Reputation is a cross-platform review and results-page audit. Then everything is written up with an owner tag on every item, developer or marketing.

What are the honest limitations?

The Mention Rate is a sampled estimate, the Presence weights are hypotheses until calibrated, and any gate we cannot measure is reported as unmeasured rather than guessed.

AI answers are non-deterministic and personalized, so the Mention Rate carries a stated run count, not a census. Published citation studies contradict each other, so the report trusts direct observation from your own live runs over any aggregate. The Presence weights stay labeled v1 until we have calibration data from thirty or more sites. Paid placements and press releases are measured but flagged as decaying tactics. Schema and llms.txt keep their near-zero treatment until evidence changes. A company too new to have a Knowledge Graph entity gets that gate marked unmeasurable, not failed.

How do you request one?

Fill in the form below with your website, an email for the report, and what you want AI to recommend you for. Free for now; we run a limited number each week.

“What you want AI to recommend you for” means the service and the city, because that is what sets the topics we query. Competitors you already know you lose to are useful but optional.