THE RECOMMENDATION LAYER · OCTOBER 22, 2026 · 8 MIN READ

What Does ChatGPT Weigh Before It Names a Firm? Three Inputs, Three Records

When ChatGPT names a competitor instead of you, it's working with what it can find. The three things it looks for, and the first move on each.

By Jacquie Baker
TL;DR: WHAT TO REMEMBER

When ChatGPT names a competitor instead of you, it hasn't sat down and compared your twenty years with their five. It can only work with what it finds. In our method, what it finds comes down to three things: facts it can confirm, other people vouching for the work, and expertise nobody else has published.

If you've put years into the work and very little into how it reads online, you're in very large company. That's the normal ratio for expert firms. It's also the fixable part.

We run these tests every month, across five engines, for firms like yours. Here's what the assistant is actually doing when a buyer asks who to hire, what we've learned about each of the three things it weighs, and the first move on each one. Most of them you can start this week.

You'll also see this work sold as AEO (answer engine optimization), GEO (generative engine optimization) or AI SEO. The labels overlap, and providers draw the boundaries differently. Our guide to AI search optimization sorts them out. We call it AI visibility.

What the Assistant Is Actually Doing

OpenAI doesn't publish how ChatGPT ranks firms. What is documented is how assistants search, and that turns out to be plenty to work with.

One buyer question rarely stays one question. Google documents this for its AI features and calls it query fan-out: the question gets split into several related searches before anything is written. Ask about a messy partnership exit and the searches touch the problem, the specialists, their reviews and the city. Each one lands on whatever record exists, and we see a similar pattern in the other engines we test.

The buyer brings their own context, too: the job, the place, whatever they've already told the assistant. That's why two people asking about the same service can get different shortlists, and why your own quick check in your own ChatGPT isn't the whole picture.

None of it works if the assistant can't reach your pages. OpenAI's crawler documentation says sites that block OAI-SearchBot won't be shown in ChatGPT's search answers. Training runs on a separate crawler, GPTBot, so you can stay visible in search without signing up to be training data.

Your first move: check that your robots.txt lets OAI-SearchBot in. OpenAI says changes take about 24 hours to apply. If you want the whole decision step by step, How does ChatGPT decide who to recommend? walks through it.

One: Facts It Can Confirm

Our method in one line: AI recommends businesses it can verify, that others vouch for, and that know something it doesn't. We call those foundation, validation and information gain. Foundation goes first, because the other two hang off it.

Foundation is boring on purpose. Who you are, what you do, where and for whom, said the same way on your site, your profiles and every directory that lists you. It's also why a smaller firm with a tidy record can get named ahead of a deeper one. Tidy is easy to read. Brilliance buried in a PDF, a podcast or a partner's head is hard to read at all.

Picture what a machine meets when it checks a firm. The website uses the full name and the business profile uses a shorter one. An old directory still has the address from two offices ago. A partner's site describes a service the firm dropped years back. None of it is a scandal. Together, it's enough doubt that the competitor whose pages all agree becomes the easier answer.

Your first move: pick one exact business name and use it everywhere. Then give each core service its own page that says what it is, who it's for and where you do it.

Two: Someone Else Vouching

Validation is what other people say about you in places you don't control: reviews, articles that mention you, association and directory profiles, the referral partner who names you on their own site. Your website can describe your expertise, but it can't vouch for it.

The vouching has to match the work. Picture ten reviews that say "great service" next to two independent pages that name you for the exact problem a buyer is asking about. The ten are lovely and genuine. The two answer the question. Buyers ask specific questions, so the evidence that does the heavy lifting is specific too.

While we're here, a distinction that trips up a lot of smart people: being cited isn't the same as being recommended. An assistant can quote your article as a source and still put a competitor's name forward for the job. A citation says your knowledge was useful. A recommendation says the assistant trusts you with the client. Why can AI cite my content and still recommend someone else? goes deeper.

Your first move: list the independent pages that would name you for your core work. If the list is short, congratulations, you've found your first project.

Three: What Only You Know

Information gain is what a page adds that the rest of the web doesn't already say. For an expert firm, it's usually the judgment that happens in conversation: how you size up a case on the first call, the question clients thank you for answering, the mistake you've watched other people make and know how to head off.

This is the one input a competitor can't reproduce, because it's your experience with your name on it, and it's the one most firms leave empty, because to the person who knows it, it feels too obvious to write down. It's also where the biggest openings are. When nobody in a category has published the real answer to a buyer's question, the first firm that does gives the assistant something nobody else can.

Written well, one of these pieces does two jobs at once. A human reader gets a real answer, and an engine gets something specific and attributable to your name. Thought leadership and the public record turn out to be the same job. Our published methodologies show how we find and structure it.

Your first move: write down the question clients thank you for answering, and the answer you give them.

Why a Small Channel Deserves Your Attention

AI referrals are still a small slice of most firms' traffic. In a Seer Interactive case study of one site, AI traffic came to 0.07% of organic search traffic, and ChatGPT visitors converted at 15.9% against 1.76% for Google organic. On that site, the few people who came through the AI door were far more likely to become customers.

That door is getting busier. In Pew Research Center's survey of February 17 to 23, 2026, 44% of US adults said they use ChatGPT, including 61% of adults aged 18 to 29. Adobe's US retail data put AI-referred traffic 393% above the same quarter a year earlier in the first quarter of 2026. Your market will move at its own pace. The direction isn't in much doubt.

How to Check Without Guessing

The tempting test is to ask ChatGPT once and take the answer personally. Don't. Answers shift by engine, by phrasing and by day, so we test the five engines buyers actually use (ChatGPT, Gemini, Perplexity, Claude and Grok) with the questions they actually ask, and read the result as a rate.

You can run a simpler version yourself: How can I check whether ChatGPT recommends my business? shows how, and How do I measure whether AI visibility work is actually working? covers what to track over time. For a scored read of your whole record, the free Recommendation Check covers 109 checkpoints.

There are three first moves on this page, one for each thing the assistant weighs. Pick the one that looks thinnest for your firm and start it this week. If you'd rather know which it is before you choose, the free check scores your whole record and shows who gets named in your place.

See where you stand

Frequently Asked Questions

What does ChatGPT weigh when a buyer asks it who to hire?

Probably Genius organizes it into three inputs: facts an assistant can confirm about a firm, independent sources vouching for it, and knowledge only that firm has published. The model's internal weights aren't public. What OpenAI does document is that ChatGPT's search answers draw on pages its crawler can reach, so the record you leave is the raw material it works from.

Why does ChatGPT name a competitor whose work isn't better than mine?

Because the name follows the record it can confirm, and a competitor's record can be clearer even when their work isn't. Consistent facts, independent mentions and published expertise are easy for a machine to read. Judgment that lives in client conversations isn't. Put yours on the record and the better work becomes the easier answer.

Can ChatGPT recommend a business whose website it can't reach?

It can still name the business from other sources, such as directories, reviews and articles that mention it. What it loses is your own site: OpenAI says sites that block OAI-SearchBot won't be shown in ChatGPT's search answers. Blocking the training crawler, GPTBot, is a separate choice and doesn't affect search. If your site blocks OAI-SearchBot, fix that first. Your own pages are where you get to tell the story.

Is being cited by ChatGPT the same as being recommended?

No. A citation means one of your pages was used as a source. A recommendation means your firm's name was put forward for the job. An assistant can cite your article and still recommend someone whose record is easier to confirm. Citations show your knowledge is useful. Recommendations also need the foundation and validation in place.

What's the difference between AEO, GEO and AI SEO?

They're overlapping market labels for work that helps a business show up in AI answers. AEO (answer engine optimization) and GEO (generative engine optimization) are the most common, and AI SEO is a looser umbrella. Providers use them differently, so compare what each one builds, maintains and measures. Our guide goes through them.

Sources

  1. Overview of OpenAI Crawlers (OpenAI). OAI-SearchBot surfaces websites in ChatGPT's search features; sites opted out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links; robots.txt changes take about 24 hours to apply for search. developers.openai.com
  2. AI features and your website (Google Search Central, updated December 10, 2025). AI Overviews and AI Mode may use query fan-out, issuing multiple related searches across subtopics and data sources to develop a response. developers.google.com
  3. How does Perplexity work? (Perplexity Help Center). Perplexity searches the internet in real time and returns conversational answers with cited sources. www.perplexity.ai
  4. Americans and AI 2026: Chatbots, Smart Devices and Views on Impact (Pew Research Center, June 17, 2026). Survey of 5,119 US adults, February 17 to 23, 2026: 44% say they use ChatGPT, including 61% of adults aged 18 to 29 (Pew notes the 2026 question wording also asked about other chatbots). www.pewresearch.org
  5. Case Study: 6 Learnings, 1 site, How Traffic from ChatGPT Converts (Seer Interactive, June 3, 2025). GA4 data from one Seer client, October 2024 to April 2025, with ChatGPT converting at 15.9% and Google organic at 1.76%, and AI traffic at 0.07% of organic. www.seerinteractive.com
  6. Adobe report: U.S. retailers see surge in AI traffic, but many websites are not entirely readable by machines (Adobe, Vivek Pandya, April 16, 2026). Adobe Analytics data covering more than 1 trillion visits to US retail sites: traffic from AI sources grew 393% year over year in January to March 2026, and in March 2026 AI traffic converted 42% better than non-AI traffic. business.adobe.com
  7. Searching the web with ChatGPT (OpenAI Help Center, checked October 30, 2026). Says ChatGPT can use location information to find local results, may use saved memories when rewriting a search query if memory is enabled, and typically rewrites a question into one or more targeted search queries. help.openai.com
WRITTEN BYJacquie Baker

Technical Co-Founder and CEO of Probably Genius, an AI visibility firm helping expert-led businesses become the named answer in AI search. Twenty-plus years of brand strategy across more than a thousand businesses, from a 16-staff Melbourne agency to Screen Australia campaigns, now focused on making experts visible, verifiable and recommendable to AI. Let's talk →

Related reading

Read the method.
Then see it run on you.

The 109-point diagnostic is this library, applied to your business. Free, about an hour to present, no obligation.

See where you stand →

A real diagnostic, not a sales call in disguise.