Being named by ChatGPT, Gemini and Perplexity
Somebody asks an assistant to recommend a business like yours. It returns two or three names and a sentence about each. There is no page two, and there are no ten blue links to be eighth in.
The short answer
AI assistants name businesses based on what your site says in plain readable HTML, how well structured that content is, and what independent sources — reviews, profiles, directories, mentions — say about you. You cannot edit the model, but every one of those inputs is something you control. The businesses being named are almost always the ones whose information is clear, consistent and machine-readable.
What we do about it
- Baseline testing: a fixed set of realistic prompts across ChatGPT, Gemini, Perplexity and Copilot, recorded verbatim
- Making sure content ships as HTML, because most assistant crawlers do not run JavaScript
- A direct answer at the top of every page, in the phrasing a person would actually use
- Structured data identifying the business, its services, its area and its hours
- Specific verifiable facts — prices, coverage, hours — because vague claims are never quoted
- Consistency between the site, the Google Business Profile and every directory
- robots.txt reviewed so AI crawlers are allowed deliberately
- Monthly re-testing of the same prompts, with the results written down
Why a business gets skipped
- The site’s content only appears after JavaScript runs
- Every page opens with company history rather than an answer
- No structured data, so nothing states what the business is or where
- The business name or suburb differs across sources, so nothing corroborates
- Almost no reviews, or reviews that all arrived in one week
- AI crawlers blocked in robots.txt by a default nobody chose
Worth knowing
When we baseline a business, the most common finding is not that assistants describe them badly. It is that they cannot describe them at all — the assistant names two competitors and does not mention the business, because there is nothing readable to draw on. That is a fixable problem, and fixing it is mostly the same work that improves ordinary search rankings.
How we run it
01
Baseline
Realistic prompts, run across the major assistants, answers recorded verbatim so there is something to compare against later.
02
Make the site readable
HTML rather than JavaScript, answers up front, structured data throughout, crawler access confirmed.
03
Corroborate
Profile, reviews and directories all agreeing. Assistants weight independent sources heavily because they cannot verify your own claims.
04
Re-test monthly
Same prompts, recorded results, plus AI referral traffic in analytics. We report what is measurable rather than what sounds impressive.
Questions about AI visibility
Can you guarantee ChatGPT will recommend us?
No, and nobody can. There is no submission form and no ranking dashboard. What is controllable is every input these systems draw on, and improving those demonstrably changes results — but treat any guarantee as a warning sign.
Which assistant matters most?
For Australian small business enquiries, ChatGPT and Google’s AI results carry the most volume today, with Perplexity growing. The good news is that the underlying work is largely the same for all of them.
How do we measure this?
Two ways. Prompt testing on a schedule, recorded so you can see change over time. And AI referral traffic in analytics, which is now a distinct and growing source you can watch.
Is this worth doing before ordinary SEO?
No. Do them together, and if you have to choose, fix the technical foundations first — they serve both. AEO is not a replacement for SEO, it is what the same work looks like when answer engines are also reading.
Related: AEO · schema and structured data · the guide version of this
Curious what the assistants currently say about you? We will run the prompts and show you.
Get a visibility check