Why does ChatGPT recommend your competitors—and what can you do about it?

You ask ChatGPT which supplier to choose. A competitor appears. Your business does not.

It is an uncomfortable result, especially if you know your offer is a good fit. But one answer cannot explain why your competitor appeared, or tell you which change would make your business appear next time.

Start with a repeatable visibility audit. Define the questions that matter, record the answers and sources, then investigate the gaps you can actually influence. This guide sets out the method we would use. It does not present an original experiment or claim to reverse-engineer ChatGPT’s ranking system.

First, establish what you are looking at

A recommendation, a brand mention and a source citation are different observations. A company may be mentioned while an independent article is cited. Your page may be cited for a definition without your company being recommended as a supplier.

Record those separately. Otherwise a report can make a modest appearance sound like a strong endorsement.

Also establish whether the answer used web search. A response that retrieves current sources is a different observation from a response without visible search activity. Save the date, product or mode used, exact question and any relevant conversation context.

Choose questions your customers would genuinely ask

Begin with the work your business wants to win. Use sales questions, objections, customer interviews and search research to build a small, useful monitoring set.

For an illustrative software business, that might include: “What should a finance team consider before buying IT cost-management software?” and “Compare approaches to allocating shared IT costs.” Add supplier questions where they reflect a real buying task.

Keep branded checks separate from non-branded questions. Asking “Why is our company the best?” tells you little about whether an unfamiliar buyer could discover you.

Our competitor keyword research process is a useful starting point for finding relevant topics. It should inform your question set rather than dictate every prompt.

Build a record you can revisit

For each observation, record:

• The exact question and why it matters commercially.
• The date, platform and visible search conditions.
• Whether your brand and selected competitors appear.
• Whether the answer recommends, mentions or compares each business.
• Which URLs are cited and what claims they support.
• Any inaccurate descriptions of your offer.
• A saved copy or screenshot of the response.

Repeat the checks under comparable conditions before drawing a conclusion. Keep alternative phrasings as separate tests. Do not keep rewriting a question until your company appears and then present that result as typical visibility.

Inspect the cited sources before changing your website

Read the pages the answer cites. Are they supplier pages, directories, publications, comparisons or explanatory articles? Does your competitor’s source explain something your own site leaves unclear?

Look for specific information: who the service suits, geography, integrations, pricing approach, implementation requirements and evidence. Record what is present and absent.

Those observations suggest areas to investigate. They do not prove that copying a cited page’s format will produce the same recommendation. A responsible audit separates “we observed this” from “we think this may be worth improving”.

Check whether search access is being restricted

Ask your technical team to review crawler access, hosting protections and the availability of important pages. OpenAI distinguishes OAI-SearchBot, used for ChatGPT search, from GPTBot, which relates to potential training use. Their settings are independent. OpenAI’s crawler documentation explains the distinction.

Do not assume allowing a crawler guarantees a citation or recommendation. It addresses an access question, not the entire selection process. Equally, do not change a company’s training preferences simply because someone has labelled all AI bots as one thing.

If the audit uncovers broader website problems, our SEO consultancy can help establish the technical priorities.

Turn observations into a focused improvement plan

Illustrative example: a buyer asks about software that works with a particular accounting platform. A competitor’s page clearly lists that integration. Your product supports it too, but the fact only appears in a sales deck.

The practical action is to verify and publish useful integration information. It is not to add “ChatGPT” repeatedly to the page or promise an AI recommendation.

Other actions might include correcting an outdated business description, improving a solution page, publishing an evidence-backed comparison or resolving a genuine accessibility issue. Assign an owner and a reason to each change.

Google similarly emphasises established SEO and helpful content for its own AI search features; it does not prescribe special AI markup. That is Google-specific guidance, not a rule for every platform. Read its guidance on AI features.

Measure progress without overstating it

Keep the original question set and record subsequent checks. If you change the set, report that separately. Track appearances and citations alongside relevant website visits and enquiries where those can be measured.

A rise in monitored visibility is a useful observation. It does not by itself establish which change caused it or how many sales resulted. Our AI visibility measurement guide explains how to keep those distinctions clear.

For evidence from client work, see the defined search and AI measures in our Serviceware case study.

Find the gap you can act on

Your next useful step is a clear diagnosis, with evidence behind each recommendation. Our AI SEO service connects that investigation with content and technical improvements. Talk to us about the questions your buyers are asking.

About the author

Molly Sishton, Founder of Re-Imagine Marketing. Molly combines marketing strategy, SEO and content expertise with hands-on experience of developing AI technology. Re-Imagine has worked with AI since 2021 and developed Sighted, its own enrichment and visibility technology. Meet the Re-Imagine team.

Recognition across our founding team and ventures: Connie Sishton featured among Wales Business Insider’s 25 Rising Stars of Welsh Business in 2025. Molly and Connie’s venture SO-ME won Best Female-Led Startup at Pitch Black Malta. Sighted was selected for Tech Nation’s Breakout 50.