Why AI recommends your competitors instead of you

What the answers say when a buyer asks for a recommendation

Method: 183 free AskedAbout checks of real local businesses (med spas, dentists, personal-injury law firms and home-services contractors) run between 2026-06-12 and 2026-09-09. Each check asks two recommendation questions on ChatGPT and Perplexity: "What are the best options for [category] in [city]?" and "I'm looking for [category] in [city]. Who do you recommend and why?", four answers per business. The check's third question names the business and is left out here, because an engine asked about a business by name almost always mentions it. Of the 178 businesses with all four answers, 107 were named in none of them and 8 in all four. A competitors-only answer names a median of six other businesses. Every answer names someone: not one of the 723 came back empty. Counted 2026-09-30 from a fresh export of the audit table; synthetic QA rows are excluded by name.

Why AI recommends you in one city and not the next one over

Ten of twelve recommendation answers named the business in its own city. Zero of twelve named it in the neighbouring city, and every one of those twelve answers named five to nine other businesses instead. Method: the six checks ran on 2026-09-30 between 13:59 and 14:01 UTC with the same two recommendation questions on ChatGPT and Perplexity, the business's own city in one check and the neighbouring city in the other; nothing else changed. Distances are between city centres, rounded. The likely mechanism, read from the sources the engines cite rather than proved by the count: an engine answers "best [category] in [city]" from pages about that city. A business that serves the next town but is only ever written about under its own city is not in the pile the engine reads for that town, so it is not named there, whatever its reviews say.

What's actually happening when AI "recommends" someone

When a buyer asks an engine for the best option in your category and city, it doesn't pull up a ranked list of links — it writes a short answer naming two or three businesses, assembled on the spot from everything it has read about your market. So when it names a competitor instead of you, that's not a verdict on who's the better business; it's a verdict on who the web describes more clearly, more consistently, and in more places. A plausible reading — a hypothesis, not something an audit proves — is that the engine names the business it can describe most confidently, and right now that's them, not you. The job isn't to become better. It's to become more legible to the machine.

The seven reasons AI names a competitor and not you

Almost every "why them and not me" plausibly comes down to one of these. They are hypotheses to check against your own measurement, not causes any audit has established; most businesses have three or four at once — which is why guessing from a single screenshot fails.

1. 1They show up in more third-party sources. Engines weigh what other sites say about a business far more than what it says about itself. If your competitor appears in "best [category] in [city]" roundups, local press, and partner sites and you don't, the engine simply has more reasons to name them and fewer to name you.
2. 2Your facts disagree across the web. If your name, category, address, or hours differ between your own site, Google, and the directories, an engine may not be able to tell which version is true. A plausible, untested explanation for being skipped is that a competitor's details line up everywhere; fixing contradictions is cheap and worth doing before testing anything else.
3. 3Your reviews live on one platform; theirs are everywhere. A strong Google profile alone often isn't enough. Whether engines weigh cross-platform review consensus is a hypothesis; if they do, a competitor reviewed decently across several sources could be named ahead of a business reviewed superbly on just one.
4. 4They're in the roundups and listicles; you're not. "Best [category] in [city]" articles do the engine's job for it. If your competitor is named in those and you aren't, you've ceded the single most citable source there is.
5. 5Your own site doesn't state the facts plainly. Engines extract facts — who you are, what you do, who you serve, where. If that's buried in marketing language or trapped in images instead of clear text (and structured data), an engine may not be able to lift it; a plausible hypothesis is that it names a business it can read more easily.
6. 6You're invisible on one engine even if you're fine on another. ChatGPT, Perplexity, Gemini, and Claude read different sources and disagree constantly. You can be named often on one and not at all on another — and most owners only ever check one. The competitor taking your slot may only be beating you on the engine you never look at.
7. 7You've never measured, so you're fixing blind. The most common reason of all: businesses react to a single screenshot. But one answer changes every time you ask — so the competitor you saw might be a fluke, and the real, repeatable gap is somewhere you haven't looked.

Why one screenshot can't tell you why

It's tempting to ask ChatGPT once, see a competitor's name, and start reacting. Don't. AI engines are nondeterministic — ask the same question twice and you'll often get a different set of businesses named (we have the run-to-run data). A competitor who appears in one answer may vanish in the next; a name you didn't see might appear most of the time. So a single screenshot tells you almost nothing about why you're losing — it's a coin flip, not a diagnosis. The only honest read is a frequency: across many runs, on each engine, how often are you named, how often is each competitor named, and which sources do the engines cite when they build the shortlist? That pattern is what tells you which of the seven reasons above is actually costing you.

How to turn "they recommended a competitor" into a fix

The diagnosis points straight at the work. Once you know which reasons apply to you, the moves are concrete and in priority order:

1. 1Find out who's actually being recommended, and how often. Not from one screenshot — from a repeated measurement across all four engines. You need the names taking your slot and your mention rate against them before you change anything.
2. 2Close the consistency gaps first. Make your name, category, location, and hours identical across your site, maps, directories, and review platforms. It's the cheapest, fastest win, and contradictions are the most common silent killer.
3. 3Get into the sources the engines cite. Find which roundups, directories, and review platforms the engines pull from for your market, and earn a presence there — those citations show what the engines read, not why they named anyone, so treat presence there as the lever most worth testing.
4. 4Make your own site machine-legible. State your facts in plain text and add Organization / LocalBusiness structured data so engines can extract who you are without guessing.
5. 5Re-measure on a cadence. Engines re-read the web on their own schedule, so any change shows up gradually if it shows up at all. Re-check the frequency with the same questions to see whether the answers moved — a before-and-after count cannot prove which change did it.

The old SEO instinct vs. what AI rewards

Reacting to one screenshot by polishing your own homepage is the most common — and most expensive — mistake we see. The competitor didn't win on their website; they won on what the rest of the web says about them.

Where AskedAbout fits

This is exactly what AskedAbout measures. The free 60-second check — your business name, category, and city — shows your AI Visibility Score (0–100), how often the engines mention you, and which competitors they recommend in your place. The $79 one-time Full Audit goes deeper: 25 buyer-intent questions across ChatGPT, Perplexity, Gemini, and Claude, each sampled 3 times — up to 300 API answers requested, with completed and failed counts shown — reporting your mention rate per engine, your share-of-voice against the specific competitors named instead, the exact sources the engines cite when they leave you out, and a prioritized fix plan. One payment, no subscription. (We query the official APIs and say so — no scraped consumer screenshots, no single-snapshot conclusions.) For context: in our checks a real local business is named in 18 % of recommendation answers about its own market, and 107 of 178 businesses were named in none of their four. If a competitor is showing up and you aren't, you are the norm, and it's fixable.

See also

• New to this? Start with the plain-English explainer: what Answer Engine Optimization (AEO) is.
• Once you know the gap, the playbook: how to show up in ChatGPT recommendations.
• The full category, side by side: AI visibility tools compared — Profound, Otterly.ai, AthenaHQ, Peec AI and Ahrefs Brand Radar vs the $79 one-time audit.
• The data behind "it changes every time": AI answers change every time you ask and which AI engine recommends your business.
• Your industry: AI visibility for dentists · for law firms · for SaaS.

FAQ