Practical guides

ChatGPT Recommends My Competitor and Not Me: What to Check

Updated August 27, 2026

You search your category in ChatGPT, it recommends three competitors, and you're not there. It's one of the most common conversations of 2026, and also one of the worst answered: most guides tell you to "create quality content," which is true and completely useless.

Here's what's worth checking, ordered by how cheap each cause is to rule out.

First: confirm they can read you

The most common cause, and the dumbest. If your robots.txt blocks AI crawlers, no optimization is possible: you can't appear in answers built from content they never read.

The four that matter today:

CrawlerWhose it isWhat happens if you block it
GPTBotOpenAIYour content doesn't enter the index ChatGPT consults
ClaudeBotAnthropicSame, for Claude
PerplexityBotPerplexityPerplexity cites sources very visibly; being left out shows
Google-ExtendedGoogleControls use for Gemini and generative answers. It does not affect your classic search ranking

This isn't always your own oversight. Many CDNs ship a "block AI bots" option enabled by default, and many templates inherit it. Check it with our free AI crawler tool: it reads your actual robots.txt and tells you which of the four gets through.

An honest nuance: blocking isn't always a mistake. If you make a living from content subscriptions, it makes sense. What you don't want is for it to be an accident.

Second: look at who it DOES recommend

Here's the information almost nobody uses. That you're absent is one data point; who appears in your place is five.

Run the search and note what kind of pages it cites:

  • Are they third-party comparisons and rankings? Then the engine isn't recommending brands, it's summarizing lists. Your path isn't optimizing your homepage, it's getting into those lists.
  • Are they competitors' product pages? Then it does recommend brands directly, and the question is what their pages have that yours don't.
  • Are they forums or reviews? The engine is weighting third-party opinion above brand material. No amount of your own copy compensates for that.

You can see that list without installing anything using our AI citations tool: you type the search and it returns the domains the engine cites.

Third: check whether your content answers the question

Generative engines cite pages that answer, not pages that present. A product page saying "the best solution for your business" answers nothing. One saying "it costs X, it's for Y, it isn't for Z" does.

Three things that correlate with getting cited, and that you can verify by looking at whoever does appear:

  1. The answer is at the top. Not after three paragraphs of introduction.
  2. There are numbers and specifics. Prices, timelines, limits, conditions. Concrete gets cited; vague gets ignored.
  3. The page is about one thing. Pages trying to cover five topics aren't the best source for any of them.

Fourth: structured data

Structured data isn't magic, and be wary of anyone selling it as such. What it does is reduce ambiguity about who you are and what you offer, which helps when the engine has to decide whether you're a relevant source.

A site with flawless schema and poor content won't get cited. One with good content and no schema can be. It's a contributing condition, not a standalone lever. You can check yours with the structured data inspector.

What nobody can promise you

Worth saying plainly, because the market is full of people promising the opposite: there is no solid evidence that any specific technique stably produces the effect of being cited across multiple engines at once. There is partial evidence, variable results, and heavy context dependence.

On top of that, generative answers are probabilistic: the same search can return different sources on two attempts, and the engine changes without notice. A one-off query is a signal, not a diagnosis.

What you can do is what any serious discipline does on uncertain ground: measure consistently, change one thing at a time, and check whether it moved. That's what AEON42 does, and it's also what you can start doing yourself today with the tools above.

Frequently asked questions

Does blocking AI crawlers affect my Google ranking?
Google-Extended doesn't: it only controls whether your content can feed Gemini and generative answers, and your classic search ranking doesn't change by blocking it. GPTBot, ClaudeBot, and PerplexityBot belong to other companies and have no influence on Google at all.
I allowed the crawlers — when will I appear?
There's no guaranteed timeline, and be wary of anyone who gives you one. Allowing access is the necessary condition, not the sufficient one: after that you depend on your content answering better than the sources cited today. It can take weeks and it may not happen.
Why does the result change every time I ask?
Because generative answers are probabilistic: the same text can produce different sources across two runs. That's why a one-off query is a signal and not a diagnosis, and why measuring several searches over time is the only thing that turns noise into a trend.
Is an llms.txt file worth anything?
It's an emerging proposal, not an approved standard, and adoption by the major engines isn't confirmed. Today it's cheap to add and does no harm, but nobody can promise you it moves the needle on its own.

Keep reading

Want to measure this on your site? AEON42 connects your Search Console and tracks your AI visibility alongside your SEO.

View plans