Why ChatGPT names your competitor — and how to check in five minutes

When a customer asks ChatGPT for the best plumber in their city — or the best CRM, dentist, or agency — the model names whoever it can retrieve as a clear, citable business. If that name is not yours, it is usually one of three things: an AI crawler never saw your pages, you are not a distinct entity in the sources the model trusts, or a competitor supplied a shorter, sourced answer. Google rankings are not enough. Google AI uses Google’s index; ChatGPT leans on Bing and its own browsing; Perplexity retrieves from a different stack. The 60-second check: ask each engine “best [your category] in [your city]” and write down who gets named. Then open /robots.txt and confirm GPTBot, ClaudeBot, and PerplexityBot are allowed.

The check (five minutes, no tools)

Run the same prompt on all three:

1. ChatGPT
2. Google AI (AI Overviews / AI Mode)
3. Perplexity

Use: best [category] in [city]. Then: who should I hire for [category] near [city]?

Record three facts per engine:

• Were you named?
• Who was named instead?
• Did the answer cite a source you control (your site, your Google Business Profile, a page you wrote)?

If you are invisible on the recommendation prompt but present on “[your brand] reviews”, you have a discovery problem, not a reputation problem. The model can talk about you when asked by name. It will not volunteer you.

Why a crawl score is not the answer a customer heard

Plenty of “AI visibility” checks stop at robots.txt, schema, and page speed. Those matter — a blocked GPTBot cannot cite a page it is not allowed to fetch.

They are not the product a buyer sees.

A buyer sees a sentence: “Here are three options in Denver.” If your competitor is in that sentence and you are not, the crawl can look healthy while the recommendation is still lost. That is the gap a live engine test is for.

The three reasons you get skipped

Crawler access. Disallow: / or a missing allow for GPTBot, ClaudeBot, PerplexityBot, and Google-Extended means the model is working off third-party pages about you — or nothing. Check https://yoursite.com/robots.txt before you rewrite a word of copy.

Weak entity. The model needs to know you are a specific business in a specific place, not a vague brand string. NAP consistency, Organization schema, Google Business Profile, and being listed in the directories your category actually uses are how you become a thing the model can name.

No citable answer. Pages that wander for 800 words before the point get summarised poorly. Lead with a 75–150 word direct answer, then evidence (numbers, named services, city, hours, what you do not do). Models prefer extractable blocks over vibe.

What to do this week

1. Unblock AI crawlers in robots.txt if they are blocked.
2. Add Organization and LocalBusiness (or SoftwareApplication) JSON-LD that matches the homepage.
3. Put a 75–150 word answer at the top of your main service page: who you serve, where, what you do, what you charge or how you work.
4. Re-run the three-engine prompt. Keep a screenshot. That is your baseline.

If you want the evidence pack instead of a checklist — who each engine named, why, and a 12-step sequence over 30 days — that is the one-time diagnostic this site sells. One payment. No dashboard subscription. Report in about five minutes.

Frequently asked questions

Is this the same as SEO?

Related, but not identical. Google AI features use Google’s index. ChatGPT relies heavily on Bing and its own browsing. Perplexity has a separate retrieval stack. Ranking on Google does not guarantee you are the name in an AI answer.

What is the fastest check I can run?

Ask ChatGPT, Google’s AI mode, and Perplexity: “best [your category] in [your city]”. Write down who they name. Then open your robots.txt and confirm GPTBot, ClaudeBot, and PerplexityBot are allowed.

Why would AI recommend a weaker competitor?

Models cite what they can retrieve as a clear entity with a short, sourced answer. A competitor with weaker service but cleaner schema, directory listings, and answer-first pages often wins the recommendation.

How is a live engine test different from an AI visibility score?

A crawl score tells you whether bots can fetch your site. A live test tells you what the model actually said to a buyer. Both matter. Customers only see the second one.