AI visibility
Does AI know enough to recommend your company?
Buyers ask ChatGPT who can help them. Check what it says about you and correct it.
By Amalgam
Reviewed by Ryan Ward5 min read
The short version
- Buyers ask AI assistants for recommendations, and the answers draw on what’s public about you.
- Clear facts, consistent descriptions, case studies with results, structured data and an llms.txt file help assistants get you right.
Buyers used to start with a search box. More of them now ask an AI assistant: who can help us build a customer portal, which firms are worth a call.
The answer can shape who they research, shortlist or contact next.
Why assistants matter to buyers
If you’re missing from an assistant’s reply, or described wrongly, you may never hear about the buyer at all.
Search is changing too. Pew Research Center found that Google users who saw an AI summary clicked a traditional search result in 8% of visits, against 15% for users who didn’t see one.
Seer Interactive data, reported by Search Engine Land, shows that when an AI Overview appears, pages cited inside it get more clicks than pages left out.
Check what assistants say about you
Ask ChatGPT and the other assistants your buyers use the questions they’d ask, in their words. Cover your category, your region and the problems you solve.
Note four things about each answer.
- Whether your company appears at all, and for which questions.
- Which competitors are recommended instead.
- What the assistant seems to understand about your products and services.
- Whether the basic facts are right.
Save each answer with the date and the exact question. Answers change between sessions and over time.
Then ask the assistant where its answer came from. If it cites pages, read them. Outdated profiles, old press releases and thin directory listings can all explain a wrong answer.
Make your basic facts easy to find
Put the basics in one obvious place on your site: what you do, who you do it for, where you’re based, when you started and who leads the company.
Write them as sentences a person would say. A line such as “We design and build billing software for regional utilities” is easier to repeat accurately than a tagline.
Make sure that text is in the page itself. Some tools that read the web for AI systems don’t run scripts, so text that loads with JavaScript can look like an empty page.
Describe yourself the same way everywhere
Your site, LinkedIn page, directory listings, partner pages and press mentions should describe you in the same terms. When they disagree, an assistant has to guess which version is current.
Pick one short description and one long one, use the same names every time, and retire old versions when your offer changes.
Publish case studies with problems and results
A recommendation needs evidence. Case studies that name the client, the problem, what you built and what changed give an assistant something to cite, and a buyer a reason to call.
Use numbers only when you can stand behind them, and say plainly when a result was qualitative.
Add structured data and an llms.txt file
Structured data describes your organization and pages in a vocabulary machines can read. Start with Organization data on your home page and Article data on your posts, and keep it truthful.
An llms.txt file is a short plain-text summary at the root of your site that points AI tools to your most useful pages. It’s cheap to publish, and few tools read it yet.
Ahrefs found an llms.txt file on 28% of the 137,000 domains using its web analytics, and 97% of those files received no traffic in May 2026. Publish one, and keep your expectations modest.
This site does the same, with Organization data on the home page, Article and breadcrumb data on each article, and an llms.txt file.
Keep checking and measuring
Repeat your assistant checks every few months and note what changed. Watch branded search and direct visits too, because a buyer who hears your name from an assistant may type it in.
