How ChatGPT Decides Which Businesses to Recommend
Updated 2026-09-01
When someone asks ChatGPT something like "what's a good [service] in [city]," the answer they get comes from one of two places: the model's own training knowledge, or — when web search is available and used — a live search of the current web, synthesized into an answer.
Training knowledge vs. live search
Without live search, the model can only draw on patterns it learned during training, which has a knowledge cutoff and skews toward businesses that were widely discussed online at that time — think well-reviewed, frequently written-about, or simply larger and older businesses. A newer or smaller business is much less likely to appear this way, through no fault of its own content.
With live search enabled, the model behaves more like a very fast, very literal researcher: it issues search queries, reads what comes back, and summarizes. This is the mode where a business's current web presence — its own site, recent reviews, directory listings, local "best of" roundups — actually has a chance to be found and cited.
What tends to get picked up in that live-search process
- Pages that clearly and specifically match the wording of the question asked
- Sources that look independent and credible — review platforms, local publications, comparison articles — rather than only the business's own marketing pages
- Recent content; a page that hasn't been touched in years is a weaker signal than one that reflects current details
- Consistent identity — the same business name, address, and description across the sites the model reads
What tends to get skipped
Thin sites with generic homepage copy and no page dedicated to the specific service asked about; businesses with no visible reviews or third-party mentions; and businesses whose online details actively conflict with each other (different addresses or phone numbers on different platforms), which makes a model less confident recommending them by name.
Checking it for your own business
The most reliable way to know where you stand is to ask the exact questions your customers ask, in a fresh chat, and read the answer critically: are you named, are you buried in a list, or are you missing entirely? AEO Score automates that check across up to five of your real customer queries and shows you the reasoning behind each result.