Why AI assistants recommend other businesses instead of yours
Because being readable by a machine is the floor, not the ceiling. Most businesses pass the technical checks and still never get named, since assistants decide whom to recommend from what the rest of the web says about you, not from your own website. I measure which of those is failing, and fix it.
Run on real assistants, with the prompts and answers recorded and dated. Nothing in a report from me is asserted without the check that produced it.
Why doesn't ChatGPT mention my business?
Almost always for one of three reasons: your site never tells a machine what you are, the facts about you disagree across the web, or nobody outside your own domain has written about you. The third is the common one, and it is the one nothing on your website can fix.
Assistants answer by pulling from sources they can cross-check. A claim that exists only on your own site is unverifiable to a model, so it gets skipped. A claim repeated on a directory, an association list or somebody else's article is safe for it to repeat.
Is this just SEO with a new name?
No, and the difference is measurable. Ranking first in Google does not get you cited: only about a third of top-ranked pages appear in the AI answer for their own query, and roughly 83% of AI citations come from pages outside the first page of results.
The mechanisms are different. Search ranks documents. Assistants retrieve passages, cross-check them against what other sources say, and then decide whom to name. You can win one and lose the other, and most businesses currently do.
What actually gets a business named?
Being written about somewhere that isn't yours. Across independent 2026 studies of tens of millions of citations, roughly 84% point at earned media rather than the brand's own site. One controlled test moved the same article from a company blog to third-party outlets and watched its citation rate go from 7.6% to 34%.
Same words, different address, four times the citations. That is why an audit that only ever looks at your website is looking at the smaller half of the problem.
What do I actually get?
A free scan gives you a score and your biggest gaps, in plain English, in about a minute. The paid audit is the one that changes anything: every check run by hand, the exact questions put to real AI assistants with the answers recorded, and the fixes written out rather than described.
The free scan measures roughly 45% of what decides whether you get recommended. It says so on its own first page. The part it cannot measure is the part that carries most of the outcome, and pretending otherwise would make the number useless to you.
How do I know you're not making this up?
Every finding names the check that produced it, and anything that could not be verified is written as unverified rather than scored. You can re-run any query in the report yourself, on the same date, and see the same answer.
I run the audit on my own site too, and publish the number even when it is unflattering. The most recent one scored 54/100 and the headline finding was that my own site is cited by nothing.
//Before you hire anyone
Four questions that tell you whether someone knows this field
This field got crowded quickly and a lot of it is vocabulary used as a smokescreen. Ask these four and most of it falls away. They work on me too.
01
Do they show you the actual answer, or only a score?
Anyone can ask an assistant a question. What matters is whether they show you the exact prompt, the exact answer, and the date. Answers shift week to week, so a score with no transcript behind it cannot be checked or repeated.
02
Are they charging you for llms.txt?
Google confirmed it does nothing for Search, and crawl studies show AI bots almost never request the file. It is free to add and it is not why anyone gets cited.
03
When they say a blocked bot makes you invisible, ask which bot.
Several of them only affect model training, not whether you appear in answers today. The two get confused constantly, sometimes deliberately.
04
Is every finding about your own website?
If nothing in the report covers what the rest of the web says about you, most of the problem was never examined.
//Proof
I run this on my own site and publish the score
The most recent audit of this domain scored 54 out of 100. Crawl access was clean, the schema was valid and the server answered in a third of a second. It was still named in zero AI answers, because everything that carries my identity lives on LinkedIn and nothing outside my own domain points here.
That is the same finding I hand most clients, and it is why the fix plan is usually mostly off-site work. If a report tells you your website is the whole problem, it has measured the easy half.
Want to know what assistants say about you?
Tell me your website and the question a customer would ask to find a business like yours. I will run it and send back what came out, along with the biggest thing standing between you and being named. No call required.