IndexerNow

AI search

My Competitor Shows Up in ChatGPT and I Don't — Why?

updated 2026-07-28
The symptom

ChatGPT or Perplexity names a direct competitor when asked to recommend a business like yours, and never mentions you.

TL;DR — the fix

This is a more useful problem than being invisible generally, because you have a worked example to study. Ask the assistant why it recommended them and where it got the information. In almost every case the answer is one of three things: they're in a directory or list you aren't, someone independent has written about them, or their pages state facts yours only imply. Each is fixable, and the assistant will tell you which one it is.

Being beaten is more informative than being invisible. When an assistant names a specific competitor, it's handing you a reference implementation — a business it considers well-enough documented to recommend. Your job is to find the difference and close it.

Interrogate the answer first

Don't start changing your website. Start by asking the assistant three follow-ups, in this order:

  1. "Where did you get that recommendation?" — names the sources that actually decided the answer.
  2. "What makes them a good fit for my question?" — reveals which facts it's relying on, which are usually more mundane than you'd guess: opening hours, service area, a specialism stated in plain words.
  3. "Do you know of [your business]?" — separates "never heard of you" from "heard of you, didn't think you fit". These are completely different problems.
The second answer is the valuable one

"Heard of you, didn't recommend you" means retrieval is working and positioning is the issue — your pages don't state the thing the question was about. That's a copywriting fix and it can be done this afternoon.

The three gaps, and what each one takes

1. They're in a source you're not

The most common finding, and the most fixable. If the assistant cites a regional directory, an industry association list, or a "best of" roundup, getting yourself into that source will do more than any amount of work on your own site. Most such listings are free or cheap and take an afternoon.

2. Someone independent wrote about them

Local press, a trade publication, a customer's blog post, a YouTube review. You can't manufacture this quickly, but you can earn it: do something worth writing about and tell the people who write about your sector. This is the slowest gap and the most durable.

3. Their pages state what yours imply

Read their homepage next to yours, out loud. Theirs probably says what they do, who for, where, and what makes them different, in the first two sentences. Yours may open with a slogan. Assistants extract claims; a slogan contains none. This one is free and immediate.

Then measure it

Run the same prompt again in a month. Not once — that's noise. Track the handful of questions a customer would actually ask and watch whether you start appearing for them. A single check tells you where you stand; repeated checks tell you whether anything you did worked, which is the only question that matters after the first week.

Check whether ChatGPT, Claude and Perplexity name your business when someone asks for one like yours. Takes a domain and about ten seconds — no account, no card.

Check my AI visibility — free

Frequently asked

Can I pay to outrank a competitor in AI answers?

No. There's no ad placement or paid inclusion in AI recommendations. Anything sold to you as guaranteed placement is not what it claims to be. The competitor is being recommended because of sources the assistant trusts, and the only route is to appear in those sources too.

My competitor is bigger. Can I actually compete here?

Often more easily than in Google. Assistants answer specific questions, and specific beats big — a business that clearly states it does one thing in one area frequently wins a narrow question against a larger, vaguer competitor. Be the obvious answer to a precise question rather than a weak answer to a broad one.

How often should I re-check?

Monthly is enough for most businesses, on a fixed set of prompts so the results are comparable. Checking daily mostly measures the model's own variability, not your progress.

Further reading

Related fixes