How to get recommended by ChatGPT and other AI assistants

How assistants decide who to name

When someone asks ChatGPT, Claude, Gemini or Perplexity for "a good web designer in Abuja" or "who does payroll software in Lagos", the answer draws on three sources.

The first is what the model learned during training. That knowledge is broad but dated, and you cannot edit it directly.

The second is the pages the assistant searches while it writes the answer. Assistants with web search switched on look up questions like these, open a handful of results and quote the ones that answer clearly. This is the part you can influence fastest, because the assistant reads your live page.

The third is what other sites say about you. Directories, review platforms, news articles and industry lists all tell the assistant whether a business is real, where it works and what it does. A business that only describes itself, on its own website, gives the assistant very little to trust.

Make your website easy to quote

Assistants quote pages that answer the question in the first lines. If your service page opens with a slogan and takes three scrolls to say what you do and where, it loses to a page that says it plainly. These changes make a website easier to read and to cite:

• Open each page with a direct answer of a sentence or two that names the business, the service and the place.
• Give every service its own page, so one page answers one question.
• Add FAQs written the way your customers actually ask.
• Show dates: when the page was published and when it was last updated.
• Serve the text as plain HTML that loads without scripts, so crawlers see what people see.
• Add structured data that states your name, services, locations and contact details, plus an llms.txt file, a proposed convention that some AI tools read for a summary of the site.
• Do not block AI crawlers in robots.txt unless you mean to.

Check your firewall and hosting settings as well. Some hosting and security services block AI crawlers by default or challenge automated visitors, which can hide a well written site from the assistants you want to reach.

Be present where assistants look

Your website is one source. The assistant wants to see the same facts confirmed elsewhere.

Claim and complete your Google Business Profile. Set up Bing Places, and verify your site in Bing Webmaster Tools so you can submit your sitemap and see how Bing indexes it. Keep a current LinkedIn company page. List the business on directories that buyers in your sector use, such as Clutch and GoodFirms for technology services.

Ask real customers for honest reviews on those platforms. Don't write them yourself and don't buy them.

Above all, keep the details identical everywhere: the same business name, the same areas served, the same phone number. When three profiles give three phone numbers, an assistant has reason to doubt all of them.

Measure, then fix what loses

Guessing wastes effort. Write down the questions your buyers ask before they choose a supplier, in their words, and put the same questions to the main assistants on a regular schedule with web search switched on.

Record each answer. Does it name you? Does it link to your site? Which other businesses does it name, and which pages does it cite? Then open the pages that win and compare them with yours. Usually the difference is plain: a direct answer at the top, specific facts, a date, a clear structure. Fix your page, run the same questions again and judge the change by what the answers say.

What we learned measuring ourselves

Elitecross measured its own visibility first. The first measurement asked Claude and Gemini the questions buyers ask before choosing a technology partner in Nigeria, without naming Elitecross in any of them. ChatGPT was not part of that first run, so this guide reports nothing about how ChatGPT answered.

Those assistants mostly cited directories and agency lists rather than individual company websites. That is why the listings above matter: when assistants lean on directories, a business missing from them is missing from the answer. It also shaped what Elitecross did next: it rebuilt elitecross.io with structured data, an llms.txt file, answer first service pages and FAQs, and it keeps measuring. The honest starting point is published on the Elitecross AI Visibility page, and the next measurements will show what changed.

What not to do

Some tactics look like shortcuts and make things worse:

• Publishing large numbers of near identical pages, one per town or keyword, with the same text and a different place name.
• Stuffing pages with keywords until they no longer read like something a person wrote.
• Writing fake reviews or paying for them.
• Blocking AI crawlers by accident through a firewall rule or a robots.txt line nobody remembers adding.

None of these builds the trust assistants look for. Most of them damage it.

What it takes to do this properly

None of the steps above is secret, and a determined owner can start on them this week. The hard part is keeping them going. Answers change from month to month as assistants update and competitors publish, so a single check tells you very little.

Doing this properly means asking dozens of buyer questions across several assistants on a schedule, recording every answer and citation, working out why a page loses, making the fix and measuring again. It also means keeping listings accurate and publishing new answer first content each month. That is steady work, and the measuring needs proper tooling rather than a spreadsheet and a spare afternoon.

If you would rather hand it over, the Elitecross AI Visibility service does this work: an audit to find the gaps, then a monthly retainer that makes the fixes, keeps listings accurate, publishes new content and reports what the assistants say. You see the measured result every month.