Guide

How to Get Cited by ChatGPT: A Practical Guide for Local Businesses

To get cited by ChatGPT, a business needs three things at once: a page that directly and completely answers the exact question a customer would type, structured data (schema markup) that tells the AI what the page is and who it's about, and enough outside corroboration, such as reviews and other sites mentioning the business, that the AI treats it as a credible source. Ranking on page one of Google does not guarantee any of this. A 2026 industry study found 77% of businesses ranking on Google's first page still weren't cited when the same query was asked of an AI model directly.

That gap exists because AI answer engines and traditional search engines are solving different problems. Google's ranking algorithm decides which links to list. An AI model deciding what to say in an answer is solving a citation problem: which one or two sources, out of everything it can see, does it actually name. Those are different questions with different winning conditions.

The two ways AI models find businesses

Tools like ChatGPT with browsing turned on, Perplexity, and Google AI Overview mostly work through live retrieval: they search the web in real time for a given question, read what comes back, and summarize it. This is the path where a brand-new, low-authority business can still win, because the model is reading current pages, not relying only on what it memorized.

The second path is parametric knowledge: what the model learned during training, before any live search happens. This only includes businesses that were mentioned often enough, across enough different sites, to get baked into the model's weights. It's much slower to influence and mostly out of any single business's direct control.

Almost everything actionable in the short term is about winning the live-retrieval path.

What to fix first

1. Answer the exact question, in the first few sentences

AI models break pages into passages and score each one for how directly it answers a specific query. A page that spends three paragraphs building up to the answer loses to a competitor's page that states it in the first two sentences. If a customer would type "does [business] offer emergency service on weekends," there should be a sentence on the site that says exactly that, not a sentence implying it.

2. Add schema markup

Schema (structured data, usually JSON-LD) is a machine-readable label for what's on a page: this is a business, here is its name, its services, its FAQ, its reviews. It doesn't change what a human sees, but it removes ambiguity for a model deciding whether a page is actually relevant. LocalBusiness, FAQPage, and Review schema are the highest-leverage starting points for most local businesses.

3. Build a real FAQ page, not a generic one

AI systems increasingly decompose a single user question into several smaller sub-questions and retrieve an answer for each one separately, a pattern often called query fan-out. A long, specific FAQ, covering pricing, service area, turnaround time, and the exact edge cases customers actually ask about, gives an AI model many small, precise passages to pull from instead of one broad page it has to interpret.

4. Get named on other sites, not just your own

Reviews, local directories, and being mentioned in a "best [service] in [city]" roundup all function as outside corroboration. A business's own website saying it's great carries little weight; other sites saying so, especially ones an AI model already trusts, carries much more. This is usually the single biggest gap between a business that gets cited and one that doesn't.

Why this compounds like SEO used to

None of these four fixes are one-time. Citation behavior shifts as models update and as competitors publish new content, so a business that fixes this once and stops usually sees the gap reopen within a few months. That's the same dynamic that made ongoing SEO work necessary for the last two decades, just applied to a newer kind of search result.