Why doesn't my company appear in ChatGPT recommendations?

You ask ChatGPT for the best provider in your category. It names three competitors and leaves you out. You go looking for a setting you missed, and you won't find one.

Companies drop out of ChatGPT answers for a handful of reasons. ChatGPT's crawler can't reach the site, or the site describes the offer too loosely for ChatGPT to match it to the question. Other websites say more about your competitors than about you. The answer can also change between one run of a prompt and the next.

ChatGPT builds each answer from the prompt, the search results it pulls at that moment and what the model already knows. Change any of those and the list changes.

How ChatGPT finds the companies it recommends

OpenAI says ChatGPT search typically rewrites your query into one or more targeted queries and sends them to its search partners. It then writes a response from the pages it retrieves and can cite them. Your prospect's exact wording is only the starting point for the search.

Take a prompt like "Which fractional marketing teams in Sydney can lead strategy and run paid media, SEO and CRM?" To answer it, ChatGPT has to match a service category, a location and a spread of capabilities, then find current sources that back each claim. If your site describes you three different ways, or makes the claim on one thin page, ChatGPT has little to work with.

Some answers draw on what the model learned in training and skip live search. Search results update as pages get crawled, while training data updates on a slower, separate timetable. Anyone who guarantees you a spot in both is guessing.

Seven reasons your company may be missing

1. ChatGPT can't retrieve your website

OpenAI's publisher guidance says any public website can appear in ChatGPT search, and asks publishers to check they aren't blocking OAI-SearchBot. Start with robots.txt. Then check your host, CDN and security plugins, because a firewall rule can block a crawler that robots.txt allows. OpenAI publishes its crawler details and IP ranges so your developer can allowlist them. A noindex tag, or key content that only loads through JavaScript, also takes your pages out of the pool.

We saw this with Sweeper.com.au, an Australian commercial cleaning equipment retailer. The site responded so slowly, and its architecture suited AI crawlers so poorly, that it was failing SEO basics before anyone looked at content. The team knew traffic was falling month on month but couldn't see why.

Crawl access makes you eligible. After that, ChatGPT picks the sources it judges most relevant and reliable, and a crawlable site with weak content still loses to a competitor with better evidence.

2. Your site doesn't connect your company to the problem

Plenty of service pages say the team "drives transformation" and stop there. A reader can't tell who you help, what you deliver, where you work or which problems you fix. A person might fill in the gaps. A retrieval system works from what the page states.

Your company name, category, services, locations, people, credentials and results should read the same on your site and on the credible profiles that mention you. Search specialists call this entity consistency, and it gives ChatGPT a clear picture of who you are.

I attended the 2026 AI Search Summit, and the category examples surprised me. I expected the biggest brands to appear in most answers, and in some categories they did. In others, the business ChatGPT named was one of the smallest and most specific in the field. A niche provider that states exactly who it serves and what it does gives ChatGPT a closer match to a detailed prompt than a household name with a vague services page.

3. Most of the evidence about you comes from you

Your website is a sales document, so ChatGPT looks elsewhere to check your claims: editorial coverage, industry directories, review platforms, professional profiles and specialist publications. You want enough independent sources that a stranger could confirm what you do without taking your word for it.

Kantar showed at the same summit that each AI platform builds answers from its own mix of brand sites, publishers, forums, social networks and information providers. AirOps analysed 21,311 brand mentions across ChatGPT, Claude and Perplexity and found 85% came from third-party sources. Your category may run differently, but if you rely on your own site alone, you're competing with a fraction of the evidence.

4. Your site doesn't answer the questions buyers ask

A prompt can read like a full brief: the best agency for a 40-person manufacturer, a set monthly budget, a HubSpot stack, a problem with lead quality. A generic services page might get you crawled, but it won't answer the comparison and fit questions that decide who gets recommended.

Write for the decision your buyer is making. Say who the service suits and when it's the wrong fit, what alternatives exist, what the engagement includes, how pricing works and what proof you have.

5. The facts about your business conflict across the web

Old directory listings, retired service descriptions, a previous office address and a LinkedIn profile for a director who left years ago all muddy the picture. The IAB's 2026 measurement framework separates hallucinations, which the platform invents, from factual inaccuracies, which come from a source that's wrong. You can trace the second kind to a page you control or can get corrected, so clean up your listings and profiles before you spend time monitoring answers.

6. Your competitors have covered the topic in more depth

One strong article won't hold a position for long. Recommendation prompts cover definitions, comparisons, costs, risks, use cases and implementation. A competitor who has published useful work across that range gives ChatGPT more evidence of expertise than a single page written to rank for one phrase. Man of Many made this case at the summit in a session titled "AI Visibility is a Topic-Level Game".

7. You tested once and trusted the result

SparkToro and Gumshoe collected 2,961 responses across ChatGPT, Claude and Google AI. ChatGPT and Google AI returned the same list of brands in fewer than one in 100 repeat runs, and the same order even less often. Across many runs, the same brands kept showing up, and that repeat presence shows you the real consideration set. Run a prompt once and you learn almost nothing about where you stand; run it 20 times and you can see which brands ChatGPT treats as the shortlist.

Case study: how Sweeper.com.au reversed its traffic decline

When Sweeper.com.au came to us, its site had almost 2,500 pages, and more than 80% of them delivered little or no traffic. The site was old, slow and losing organic traffic each month. The business needed to recover its search rankings and start appearing in AI recommendations, and the existing site couldn't support either.

We rebuilt the site from the ground up. The work covered:

• Technical foundations. We fixed site speed and built an architecture that search and AI crawlers can load and read.
• Structure. We removed a large share of the pages that weren't earning traffic and reorganised the rest, which lifted the overall site health score.
• Schema. We added structured data that matches the content on each page.
• Content. We rewrote the copy on the pages we kept and added richer content that answers buyers' questions.

After launch, the organic decline reversed. New unique visitors rose 29%. Sweeper also recorded visits from AI platforms for the first time, and some of those visitors became leads.

The rebuild touched four of the seven reasons above: crawl access, clarity, answering buyer questions and topic depth. Cutting 80% of the pages made the site stronger, because the pages that remained gave ChatGPT and Google a clearer picture of what Sweeper sells.

Try this: a 15-minute AI visibility check

You can get a quick read on where you stand before you spend anything. Set aside 15 minutes and a notepad.

• Write your buyer's question. Think of the question your best customer would type into ChatGPT when looking for what you sell. Include the details they would: location, business size, budget or the problem they need solved.
• Ask it three times. Open ChatGPT in a private browser window so your history doesn't shape the answer. Ask the question, then start a new chat and ask it twice more.
• Log each answer. Note which businesses appear, where you appear if at all, and which sources ChatGPT cites.
• Study the winner. Pick the competitor that appears most often and open the sources ChatGPT cited for it. Count how many are the competitor's own pages and how many are directories, reviews or articles.
• Compare. Search for your own business in those same directories and publications. Each place your competitor appears and you don't goes on your list.

Three runs gives you a quick read on who ChatGPT favours and why. For a figure you can report on, the IAB suggests at least 50 queries with several responses each.

What to fix first

Work through these in order. The early steps are quick and everything after them depends on them.

• Access. Confirm Google, Bing and OAI-SearchBot can crawl and index your important pages. This is often a one-day fix.
• Clarity. State your category, audience, services, locations and point of difference in the same words across the site.
• Consistency. Correct outdated details on the major directories, profiles and review platforms.
• Evidence. Publish named expertise, case studies, original data, methodology and dates.
• Depth. Build content around the buyer's whole decision, covering fit, cost, risk and alternatives.
• Third-party coverage. Earn reviews and mentions in publications with real editorial or industry standing. This takes the longest, so start it alongside step 1.
• Measurement. Track a fixed set of prompts across several platforms each month and keep the prompts stable so you can compare results.

How should AI visibility be measured?

The IAB Measuring Visibility in the AI Era framework splits brand measurement into four layers: presence, prominence, portrayal and persuasion. In practice you're checking whether the brand appears, where it sits in the answer, how accurately ChatGPT describes it and whether any of that leads to action.

The IAB also notes that AI outputs vary from run to run. It treats a program of fewer than 50 queries as exploratory, and recommends several responses per query, prompts that cover different intents and open reporting of uncertainty. If a dashboard shows you a single "ChatGPT ranking" and won't tell you the prompts, platforms, repetitions and collection method behind it, ask for them before you trust the number.

Where not to spend your budget

You'll find no evidence that an llms.txt file, an AI-only content format or a stack of formulaic FAQs makes a company recommendable. Google states that you don't need new machine-readable files, AI text files or markup to appear in AI Overviews or AI Mode, and that you don't need any special schema. OpenAI's main ask is access for OAI-SearchBot. Structured data still helps search engines read your content when it matches what's on the page, though it won't make up for thin evidence or an unclear offer.

Where to start

If ChatGPT leaves you out, diagnose the problem before you commission more content. Check crawl access and indexation. Run your core prompts across several platforms and log who appears and how ChatGPT describes them. Look at the sources ChatGPT cites for the competitors it recommends, compare their evidence with yours, and work on the biggest gap first.

How Purple People can help

We help growing Australian businesses connect search, answer engine optimisation, content, digital PR and website strategy. We'll audit how your company appears across AI platforms, find the gaps in your own and third-party evidence, and turn what we find into a content and authority plan your team can act on. For Sweeper.com.au, that work reversed a traffic decline and brought in the business's first AI-referred leads. Explore our organic visibility service or marketing strategy service.

Sources and further reading

OpenAI: ChatGPT search

OpenAI: Publishers and Developers FAQ

OpenAI: Overview of OpenAI crawlers

Google Search Central: AI features and your website

IAB: Measuring Visibility in the AI Era

SparkToro: AI recommendation consistency research

AirOps: The influence of offsite signals in AI search

Kantar: Chosen by AI, The New Rules for Winning in AI Discovery. Presented at the 2026 AI Search Summit.

Man of Many: AI Visibility is a Topic-Level Game. Presented at the 2026 AI Search Summit, July 2026.

Get an AI Visibility Audit See Our Work