ChatGPT.

We get your business surfaced inside ChatGPT answers — not just when someone asks for you by name, but when they ask the question that should lead to you. That means engineering your site, schema, and citations so OpenAI's retrieval layer treats you as a high-trust source, and so the model's training data treats you as a category-defining brand. We track every prompt that should mention you, log the answers ChatGPT actually returns (with screenshots), and tune until you're in the shortlist.

ChatGPT is the front door. When someone is evaluating tools, services, or providers, they open ChatGPT before they open Google. The model returns two to four named recommendations per question — that's it. If you're not in that shortlist, you don't get the consideration. SEO got you to page one of Google. AEO gets you into the answer itself. ChatGPT's answer is the new page one, and most categories already have entrenched defaults that the model keeps repeating.

ChatGPT is half-trained, half-live. The model has a baked-in worldview from its training cutoff — that's where its default opinions come from — but Search is layered on top through a Bing-backed retrieval system. So winning ChatGPT is two jobs at once: (1) becoming a name the base model recognizes (mentions across high-authority third-party content) and (2) becoming a source the live retrieval ranks well (clean schema, fast pages, dense entity coverage on your site). Most competitors only do one of these. We do both.

If you want to win ChatGPT you need to know how it actually picks what to say. Here's the short version of what's happening every time a user types a prompt.

1. 01

   Retrieval layer

   When ChatGPT decides it needs fresh information, it calls a Bing-powered web index. The query it sends isn't your prompt verbatim — it's a rewritten search string. Ranking well for those rewritten queries is the entry point.

2. 02

   Source preferences

   ChatGPT's retrieval favors structured pages with clear headings, FAQ schema, and direct factual answers near the top. It also weights third-party validation — review sites, industry press, and aggregators with strong authority.

3. 03

   Citation behavior

   Search-mode ChatGPT cites with inline links and a sources rail. Non-search answers cite from memory — meaning you only show up if the base model already knows you. The fix for the second is consistent off-site mentions over time.

4. 04

   Training signal

   OpenAI re-trains on a moving window of public web data. Brands that get mentioned in independent, well-cited contexts (Reddit threads, Wirecutter-style roundups, Wikipedia) become part of the next model's default vocabulary.

5. 05

   Refresh cadence

   Live retrieval updates within hours of a crawl. Training-derived knowledge updates only when OpenAI ships a new model checkpoint — typically every 6 to 18 months. We optimize on both clocks.

No silver bullets. These are the levers that move the needle in ChatGPT specifically, ranked by what tends to matter most for our clients.

• —Entity schema (Organization, Service, FAQPage, Review) — clean and complete
• —Direct-answer paragraphs in the first 150 words of each page
• —Comparison and 'best [category]' content with you placed by name
• —Third-party seeding: Reddit, niche forums, press, citations on authority sites
• —Branded search lift tracking — when ChatGPT mentions you, Google traffic for your brand spikes within 48 hours, and we measure it