Why does AI recommend your competitor and not you?

It is not the business with the best website. If a competitor keeps being named and you aren’t, it’s almost always because more pages on the wider web describe them for that question, and those pages are easier for the assistant to read and check. That’s fixable, but not by rewriting your homepage.

How do AI assistants decide which businesses to name?

When you ask ChatGPT, Claude, Gemini, Perplexity or Google’s AI Overviews for a recommendation, the assistant usually runs one or more web searches in the background, reads the pages it finds, and writes an answer from them. The businesses that get named are the ones that appear, described in a similar way, across several of those pages.

There are three steps, and each one can exclude you. First the assistant retrieves pages, using a search index: Google’s for Gemini, AI Overviews and AI Mode, OpenAI’s own crawl for ChatGPT, Perplexity’s own index, and a live web search for Claude. If your business isn’t on the pages that come back for that question, you’re out before anything else happens. Second, it reads what it retrieved and pulls out the parts that answer the question. Pages that state things plainly get used; pages that hint get skipped. Third, it decides what it’s confident enough to say. A business named on one page is a maybe. A business named on five pages, described the same way each time, is a recommendation.

Google’s documentation says its AI features draw on the same ranking systems as normal search, and OpenAI describes ChatGPT search the same way: a search, then an answer written from the results. There’s no separate list of approved businesses. There’s the web, read quickly.

Why is my competitor mentioned when we’re not?

Usually one of three things, and often all three. The sources the assistant found for that question mention your competitor and not you, your competitor has a page that answers the question in words the assistant can lift, and your competitor is described consistently wherever they appear.

Coverage is the big one. For a question like “who’s a good payroll provider for a 50-person company”, the assistant is reading comparison articles, directories, review sites, trade press, forum threads and LinkedIn posts. If your competitor is in those and you aren’t, the answer is written before your website is ever opened.

Match is the second. Your competitor may have a page titled with the customer’s question that answers it in the first paragraph. Your equivalent may be a services page that describes what you do in general terms. The assistant is looking for the former.

Consistency is the third. If you’re “Smith & Co” on your site, “Smith and Company Accountants” in a directory and “Smiths” on LinkedIn, with two addresses and three descriptions of what you specialise in, the assistant can’t place you confidently, so it doesn’t.

Is it because their website is better than ours?

Mostly no. A better website helps you get cited for specific questions and helps the assistant confirm facts about you, but a recommendation for a “who should we use” question is built from what other sites say. Two businesses with equally good websites can have completely different visibility depending on how often, and how consistently, the rest of the web describes them.

That’s the part most businesses miss. They rewrite their own pages, see no change, and conclude AI search is random. It isn’t random. It’s reading sources you haven’t looked at. The source ecosystem in AI search explains which ones tend to matter.

Which sources are the assistants using?

The mix varies by assistant, but the usual ones are comparison and “best of” articles, directories and review platforms, industry publications, LinkedIn, YouTube, Reddit and other forums, Wikipedia for larger organisations, and the businesses’ own sites. Google’s assistants lean towards what ranks well in Google, including video and forum threads. ChatGPT leans more on pages it can read directly.

You don’t need to guess. Ask the assistant the question where your competitor is named, then look at the sources. Google AI Mode and Perplexity show links beside the answer. ChatGPT shows citations when it has searched the web, and you can ask it directly which sources it used. Claude does the same. Note every page cited, then check whether you appear on any of them. That list is your work plan, and it’s usually shorter than you’d expect.

Does the answer change from one assistant to the next?

Yes, often a lot. The same question can name your competitor in Google AI Overviews and not in ChatGPT, or the other way round, because each assistant searches a different index and weighs sources differently. Before you conclude you’re invisible, check all of them. How to check what ChatGPT says about your company covers how to do that in an hour without fooling yourself.

What can we do about it?

Five things, in this order. Write down the questions where your competitor is named and you aren’t. Find the sources cited for those questions. Get described on those sources. Make one page on your site answer each question directly. Then measure it for weeks, not once.

Getting described on the right sources means the plain routes: being listed in the directories and comparison pieces that are cited, being covered by the publications the assistants read, answering the forum threads as yourself, and having customers review you where the assistant looks. None of it is clever. It’s the coverage your competitor already has.

For your own pages, open with the answer, use the words the question uses, and be specific. The GEO study from Princeton and partner universities found that adding statistics, quotations and cited sources to a page increased how often it was used in AI answers, while adding keywords didn’t help. What pages to fix first for AI search shows how to choose where to start.

Then measure. Ask the same questions across the same assistants every week and record who’s named. How competitor benchmarking works in AI search covers how to do that fairly. If a smaller rival with a thinner site keeps beating you, why smaller competitors beat you in AI answers explains what’s going on.

Where to go from here

This is what our monthly plans are built for. We track the questions your customers ask across ChatGPT, Claude, Gemini, Perplexity and Google’s AI Overviews every day, show you where competitors are being named instead of you and which sources are doing it, and turn the gaps into the content we write each month. Are you free for 20 minutes in the next week or so? We’ll bring your data to the first call.