Customers increasingly skip the search-results page entirely. They ask an assistant a question — "who's the best near me?" — and the assistant answers with two or three business names. If yours isn't one of them, that customer never knew you existed. There is no page two.
AI assistants recommend what is published where a machine can read it — not what is true. That single sentence explains almost every result we've measured.
The cleanest proof from our own audits: a dental practice whose team genuinely speaks five languages was named #1 by ChatGPT and Grok for a language-specific search — because two insurance and doctor directories had published the languages. Two other assistants missed the practice entirely, because they read the practice's own website and Google listing, where the languages appeared nowhere. Meanwhile a competitor won the general multilingual query outright by stating "English, Spanish, and Arabic" in one plain sentence on its homepage.
Three published languages beat five unpublished ones. The gap is almost never capability. It is publication.
We ask real customer questions in the consumer AI products themselves — signed out, with memory off, so nothing personalizes the result — and record whether the business is named, who was recommended instead, and which sources each assistant cited. The answers score four weighted pillars, producing an AI Referral Score from 0–100.
Is the business named when customers ask without using its name? Won substantially on third-party sources the assistants cite.
What do assistants say when they do name you — and do your reviews and ratings support or undercut it?
Can a machine read the site? Content in the served HTML, question-formatted sections, structured data, visible facts.
Do the name, phone, address, and facts match across every profile an assistant might cite?
A full engagement moves through four phases: a baseline audit (the question battery, recorded and scored), a prioritized action plan, the implementation — machine-readable structured data in the site's served HTML, direct answers under question-formatted headings, a page for every service, corrected and connected business listings, review-profile work, and explicit access for AI crawlers — and a re-test of the identical questions at 60–90 days, delivered side by side against the baseline so the change is measured, not asserted.
Two things we say before any engagement: no one can guarantee a particular AI answer — these are third-party platforms that change without notice. And a perfect website alone is not enough: Discovery, the heaviest-weighted pillar, is won substantially on third-party sources beyond the website. The program works both.
AI Referral Optimization is the practice of making a business the answer when customers ask an AI assistant — ChatGPT, Gemini, Grok, Perplexity, or Claude — for a recommendation. It measures how each assistant currently answers real customer questions, then improves the published, machine-readable information those assistants rely on: the business website, its structured data, and the third-party profiles and directories the assistants cite.
SEO competes for a ranked position on a results page a human then scans. AI Referral Optimization competes for the answer itself: an assistant names two or three businesses and most customers never see a results page at all. The work overlaps with SEO but adds AI-specific requirements — content readable without JavaScript, question-formatted sections with direct answers, structured data served in the page HTML, consistency across the third-party sources assistants cite, and explicit permission for AI crawlers.
Because assistants recommend what is published where a machine can read it — not what is true. A real capability that appears nowhere in crawlable text does not exist to an AI. In our audits, a practice whose team speaks five languages lost a language query to a competitor that had published three languages in one sentence on its homepage. The gap is almost never capability; it is publication.
We ask real customer questions in the consumer AI products themselves — signed out, with memory off, so no personalization skews the result — and record whether the business is named, who is recommended instead, and which sources each assistant cited. A full baseline runs 22 questions across five platforms (110 recorded answers) and scores four pillars: Discovery, Reputation, Website Readiness, and Listing Consistency, producing an AI Referral Score from 0 to 100.
No — and no honest provider can. These are third-party platforms that change their retrieval systems without notice. What we deliver is measured: a baseline audit, a prioritized action plan, the implementation, and a re-test of the identical questions at 60 to 90 days so you can see exactly what moved.
Expect little visible movement in the first month — published changes must be crawled and re-indexed before assistants pick them up. The formal re-check runs 60 to 90 days after implementation, comparing the identical questions on the same platforms against the baseline.
Yes. The free AI Referral Audit asks five real customer questions across ChatGPT, Gemini, Grok, Perplexity, and Claude, records every answer, and returns a plain-English report with a preliminary AI Referral Score — where you're strong, where you're invisible, and which competitors are winning the referrals that should be yours. There is no cost and no obligation.
24 to 48 hours. You submit your name, email, and website, confirm the email address so we know the report will reach you, and complete a two-minute form. We then run the five questions across all five platforms and send your report.
Yes. The report is emailed to you and it's yours to keep whether or not you ever hire us. We show you what the full AI Referral Optimization program includes alongside your results, so you can see the difference — but there's no obligation either way.
The free audit is a sample: five questions, one preliminary score, no implementation. The full program asks 22 questions across the same five platforms, scores four measured pillars, names the competitors taking your answers, implements a prioritized action plan across your website and listings, and re-tests the identical questions at 60 to 90 days. It runs $1,900 to $2,400 depending on the size and complexity of your website.
The free AI Referral Audit asks five real customer questions across ChatGPT, Gemini, Grok, Perplexity, and Claude — and sends you a plain-English report with a preliminary AI Referral Score. No obligation.
Get My Free AI Referral Audit →