Why your AI competitors are not who you think
The businesses an assistant names in your place are your real competitors in AI search, and they are often not the ones you compete with elsewhere. A local firm may be named beside a regional one you never thought about. A store may be named beside review sites and marketplaces, not its usual rivals. A small specialist may be named ahead of a bigger brand.
That is the reason to look. The assistants build their answers from what is written about businesses across the web, so the list that comes out reflects who is well documented by other people, not who has the biggest sales or ad spend. Our write-up of how AI assistants pick businesses covers the field test behind this idea, so we will not retell it here.
This page is the competitor half of an AI visibility check. If you have not yet run the questions, begin with the guide to running your first check. Here we assume you have a set of answers and want to learn from the names in them.
Record every business named, open the pages the assistant drew on, compare them with what you have, and pick one real gap to close.
The method in four steps
The diagram above lays out the sequence. In words, it goes like this.
- Record who is named. Every business, for every question, in the order the assistant gave them.
- Open the sources behind them. Where the assistant shows its sources, open the pages. Where it does not, search for the named business and see what appears.
- Compare with what you have. For each kind of source that supports the others, ask whether anything equivalent exists for you.
- Choose one gap. Pick the most realistic difference you can actually close, and do that first.
The common failure is stopping at step one. A list of names feels like a finding. It is only an inventory. The useful part is why those names, and that lives in the sources.
You will want a place to keep all this, and the layout in an AI visibility tracking spreadsheet has columns for "Others named" and "Sources shown" for exactly that reason. The sections below take the steps one at a time.
Step one: build the list of who gets named
After a run of your question set, go through your answers and pull together every business that was named. Then count how many answers each appeared in.
A short table in your notes is enough, with these columns.
- Business name.
- Answers it appeared in, as a count out of the total.
- Which assistants named it. One, two, three or four.
- Which kinds of question. Shortlist, comparison, place, doubt.
- Typical wording. The usual reason given.
Sort the list by count. You will probably see a pattern. A few businesses appear everywhere. A larger group appears now and then. Many appear once.
The ones that appear everywhere, across several assistants and several kinds of question, are your reliable competitors in AI answers. The ones that appear once are mostly noise, and you can set them aside. Assistants vary from run to run, so a single appearance proves little. Our page on why AI answers change explains the variation.
Leave marketplaces, directories and review sites in a separate column. They are sources, and sometimes also named as places to go. They matter, but they are a different kind of competitor.
Step two: open the sources behind the names
This is the step that makes the check worth doing. For each of your reliable competitors, find out what the assistant is leaning on.
Where the assistant lists sources, as some do, open each link and note what kind of page it is. Where it does not, do a normal search for the competitor's name plus the topic of the question, and see what turns up on the first pages. That is a reasonable stand-in for what the assistant could have found.
Make a note for each source of:
- Type. Their own site, a review site, a forum thread, a news article, a directory listing, a guide.
- Who wrote it. The business itself or someone else.
- What it says. The reason someone gave for choosing them.
- How old it is. Recent or years old.
The "who wrote it" note matters most. A business describing itself is a claim. Someone else describing it is closer to an endorsement. Assistants appear to lean on the second kind, which fits what we describe in what AI engines cite.
You may find it takes ten or fifteen minutes per competitor. That is fine. Do it for the top three or four, and no more.
Step three: compare what they have with what you have
With the sources in front of you, go down this checklist for each competitor and answer honestly for yourself.
Reviews. Do they have more reviews, more recent ones, or reviews that say specific things? Our page on reviews and AI recommendations covers why specifics matter.
Third-party mentions. Are they listed in guides, roundups, local articles or association pages that you are not in?
Discussion. Are they talked about in forums and communities? See Reddit and AI search for why these threads carry weight.
Clear pages of their own. Do their pages state, in the first lines, who they serve, what they do and where? Is there a page for each service or product line?
Consistent facts. Do their name, address, hours and descriptions match across sites?
Depth. Do they have buying guides, how-to articles or comparisons that answer the questions in your set?
Mark each item as "they have it, I do not," "we both do," or "I have it, they do not." The first group is your list of gaps. The third is worth noting too, because it shows strengths that are not yet being recognized.
Which gaps are fast and which are slow
Not every gap is equal. Some you can close in a week, and some take a year. Sort what you found.
Fast. Pages that do not exist yet on your own site. A service page, a comparison, a guide that answers one of the questions. These are in your control and can be done quickly. Inconsistent business details across listings are also quick to fix.
Medium. Reviews. You cannot buy them honestly, but you can ask satisfied customers more systematically, and that builds over weeks and months.
Slow. Being written about by other people. A mention in a local article, a guide, or a respected community thread comes from relationships and from being worth mentioning. It is the gap with the largest effect on whether you get named, and the one nobody can shortcut.
Be wary of anything that promises to close the slow gap quickly. Bought placements in low-quality places do not tend to be cited. The aim is to be written about because you did something worth writing about.
For a page-by-page view of what to do after finding gaps, see what to do after an AI visibility check.
Do not copy their pages
It is tempting, once you have read a competitor's pages, to write something similar. Resist it for two reasons.
The first is practical. A near copy gives an assistant no reason to prefer you. It has a page like that already, from someone who also has the reviews and the mentions. A copy adds nothing to what the assistant can say.
The second is about substance. What helps a business get named is usually something true and specific about it that other sources can confirm. A particular service area. A specialty. A way of working. A page that says what is actually different about you, with specific facts, can become the answer to a question that a generic page never will.
So use their pages as a map of the question, not as a template for the answer. They show you which questions the assistants are answering and what a satisfactory reply looks like. Then write the reply that is true for you.
We do not name competitors in our own pages and we would suggest you do not publish comparisons that run them down. A neutral, factual comparison page that helps a buyer choose is fine, and often useful. A hit piece is not, and it tends to read that way to assistants and to people.
AI competitors against ordinary search competitors
Your rivals in Google results and your rivals in AI answers overlap, but not completely. Compare the two lists.
On both lists. The strongest competitors. They are well documented and well ranked.
In AI answers only. Businesses with a lot written about them by others but weaker websites. They show that third-party mentions can carry a business a long way.
In Google results only. Businesses with strong pages and rankings but little discussion elsewhere. They may be vulnerable in AI answers.
That comparison tells you where your own efforts can go. If you already rank well but are missing from AI answers, the gap is probably about what others say about you. If you are named in AI answers but rank poorly, the gap is probably on your own pages and links.
Search is changing, and ordinary search remains important. People call the work of earning mentions in AI answers GEO, short for generative engine optimization, or AEO, short for answer engine optimization. Whatever it is called, it sits beside classic search work and does not replace it. Our page on SEO, GEO and AEO compared sets out how the three relate.
Keep the competitor list current
Competitor lists change. A new business gets some good press and appears in a few answers. An old one fades. A month's reading is a reading, not a fact.
Repeat the competitor tally each time you run your question set. It is quick if you kept the "Others named" column. After three runs you can tell which names are stable and which come and go.
Also keep a short log of what you changed and what changed around you. If a competitor suddenly appears in more answers, check whether they published something or were mentioned somewhere new. That is how you learn what moves the needle in your category, using other people's experiments.
If you want to compare yourself against a whole category rather than a few names, our page on the AI visibility benchmark covers picking the comparison set.
The names tell you who the assistants know. The sources tell you why. Close the one gap you can actually close, then run the check again.
A worked example of reading one competitor
Suppose a small accounting firm runs its question set and finds that one other firm appears in thirteen of forty answers, in all four assistants. That is a reliable competitor, so the owner studies it.
She opens the sources in the answers where that firm appeared. Four are review pages with several dozen reviews, many of which mention the same specialty, small construction businesses. Two are local articles, one a chamber of commerce profile and one a roundup of regional advisors. One is a discussion thread where a contractor asked for a recommendation and a neighbor named the firm. The firm's own site appears once.
Now she compares. Her own firm has a handful of reviews, none naming a specialty. She is in no roundup. She has a single services page, and it opens with a mission statement rather than who she serves.
The gaps sort easily. Fast: a page that says in its first lines who she serves, with a section per kind of client. Medium: a routine for asking clients for reviews that say what the work was. Slow: earning a mention in the sort of roundup the assistants used. She cannot control that, but she can decide to be the sort of firm a writer would include, and say yes when one asks.
She chooses one gap, the services page, and does it that week. Next month she runs the same questions and looks at both counts again.
Where RunOctopus fits in a competitor check
RunOctopus does not offer a free competitor checker for AI answers. The method above is the free way to do it, and it works.
What RunOctopus does is the building that closes the fast and medium gaps. It is a done-for-you service. Ollie, our AI, researches a business and writes and publishes pages on its own website, with a team that plans, writes, reviews and publishes alongside it. It works with Shopify, WordPress, WooCommerce and Wix, and the owner installs nothing and approves nothing. After the first week of 8 long-form articles, 6 pages and a custom interactive tool for the niche, it continues at 175 or more pieces a month.
Paying customers also get reports on where ChatGPT, Perplexity, Claude and Gemini name their business. We cannot make other people write about you, and we will not claim to. We can make sure your own pages answer the questions clearly.
Florida Realtor Careers went from 8 to 116 Google clicks a month since launching in June, per Google Search Console for the 28 days to September 29, 2026.
The free site read reads your address and shows the plan, with no account and no call. It does not check AI answers.