How AI engines choose which brands to name.
An AI answer names a handful of brands and leaves out the rest. This chapter covers where those names come from: what the model remembers, what it looks up while it answers, and the four things that decide whether yours is one of them.
Chapter 1 of 6 of the AI visibility guide
What changed when buyers started asking AI?
A search engine returns a page of links and lets the buyer choose. An AI assistant writes one answer and names a handful of brands in it. There is no page two: a brand that is not in the answer is not considered.
The answer is built from sources the buyer rarely opens. When Proofsource asked 25 buyer questions in Tesla's markets on four engines (Tesla, public AI answers, Oct 4, 2026), the 100 answers cited 455 different websites, and fewer than 2 in every 100 citations went to tesla.com. Even for a brand everyone knows, its own site is one voice among hundreds.
Clicks follow the answer. Pew Research Center tracked 68,879 Google searches by 900 US adults in March 2025: when an AI summary appeared, people clicked a regular result 8% of the time against 15% without one, and clicked a link inside the summary in 1% of visits (our write-up of the Pew study). Being named in the sentence matters more than being one of the links under it.
ChatGPT answers that named the brand when the model ran no web search, against those where it searched.
215 ChatGPT answers, Aug 22 to Oct 1, 2026Brands ChatGPT put in its first search were named that often; brands in none of its searches, 7%.
2,502 ChatGPT answers, Oct 4 to 13, 2026When any of four engines named the brand for a question, all four did this often.
292 questions, Sep 30 to Oct 4, 2026Where do the names in an answer come from?
Two places. One is what the model learned in training, which stops at its knowledge cutoff and changes only when a new model ships. The other is what it looks up on the web while it answers. A page you publish this month can reach the second, never the first, until a model trained after it is released.
Whether the engine looked anything up was the biggest single difference we measured. Of 215 ChatGPT answers we scored, 147 ran no web search at all; those named the brand 6% of the time, and the ones that searched named it 78% of the time. Comparing the same questions against themselves narrowed the gap to 24% against 78%, still far apart (how we measured it). OpenAI's API guide says the model can choose to search or not based on the prompt, so one percentage mixes answers that looked and left you out with answers that never looked.
When ChatGPT does search, it writes its own query. OpenAI's help page says ChatGPT typically rewrites a question into one or more targeted queries, and in our sample those queries often already held brand names: a brand in ChatGPT's first search was named in 89% of answers, a brand in no search in 7% (the full breakdown). The shortlist is mostly settled before the search runs.
Google goes further. Its guidance for site owners says AI Overviews and AI Mode may use a query fan-out, issuing several related searches across subtopics to build one response (what that breaks in tracking). The question being answered is not only the one the buyer typed.
Do the four engines agree on which brands to name?
Less than their scores suggest. We put the same 292 buyer questions to ChatGPT, Google AI Overviews, Perplexity and Claude on the same days. ChatGPT, AI Overviews and Claude each named the brand on 22% to 23% of question days, which looks interchangeable on a dashboard. They named it on different questions: when any engine named the brand, all four did only 32% of the time (the agreement study).
So track each engine on its own, and read how each one chooses before you work on it: ChatGPT, Perplexity, Claude and Google AI Overviews.
What decides whether an AI engine names your brand?
Four things, and they gate each other in this order. Content work cannot help if a crawler cannot read the page, and a perfect page cannot help if the engine thinks you are a different company.
Can its crawlers read you?
The bots each engine sends must reach your pages and see real text, not an empty shell.
Does it know who you are?
The engine has to tell your company apart from others with similar names.
Are you on the pages it cites?
Comparison pages, review sites and lists decide most shortlists. You need to be on them.
And a fourth that runs through all three: answer the questions buyers actually ask.
Crawlers and identity are chapter 4, cited pages are chapter 5, and the questions start in chapter 2.
Keep reading
Every chapter
Related reading
- Pew Research Center: Google users are less likely to click on links when an AI summary appears (July 22, 2025) www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/
- OpenAI Help Center: Searching the web with ChatGPT help.openai.com/en/articles/9237897-searching-the-web-with-chatgpt
- OpenAI API documentation: Web search tool guide developers.openai.com/api/docs/guides/tools-web-search
- Google Search Central: AI features and your website developers.google.com/search/docs/appearance/ai-features
Proofsource figures link to the post or study that measured them, with its dates and method. Tell us if anything here is out of date.
Find where AI leaves you out. And fix it.
See which buyer questions name your brand, who gets named instead, and what to fix. 200 answers. 25 Prompts/Question. Top AI engines. 2 days. Free. No card.
Google AI OverviewsGoogle AI Mode
