ChatGPT doesn't show your link.
It tells your buyer about you.
Or your competitor.

ChatGPT SEO is the practice of optimizing your website, brand, and external citations so that ChatGPT surfaces and cites your business when users ask category-relevant questions. ChatGPT pulls from its training corpus (which is updated periodically) and from real-time web browsing (which uses Bing's index plus a curated set of trusted sources). Optimizing for ChatGPT therefore means optimizing for both — strong baseline Bing visibility plus the structural and authority signals that train OpenAI's models to associate your brand with a topic.

Both, depending on the mode. The standard chat mode often answers without citations from training data alone. ChatGPT Search and the web-browsing mode cite sources by default — and those citations link directly to source pages, driving real click-through traffic. Even when ChatGPT answers without a visible citation, your brand can be mentioned by name inside the response, which influences buyer decisions before they ever click anything. Both surfaces matter; we optimize for both.

Three layers. Layer one: training-corpus authority — domains with high overall authority (Wikipedia, major publications, established industry sources) get weighted heavily during model training, which is why a single mention in a tier-1 publication can drive citations for years. Layer two: real-time retrieval — when ChatGPT browses, it uses Bing's index plus OpenAI's curated source lists, weighted toward recency and entity-relevance. Layer three: entity disambiguation — when multiple businesses share a similar name or topic, ChatGPT picks the entity with the strongest schema signals, sameAs links, and topical clustering across the open web.

Because ChatGPT's web-browsing mode runs queries against Bing's index. If you don't rank in Bing for a buyer's question, ChatGPT's web mode can't find you in real time. This is the single most-overlooked tactical lever in ChatGPT SEO: most agencies stopped paying attention to Bing a decade ago, but Bing is now the retrieval layer for the largest generative engine in the world. We submit and verify Bing Webmaster Tools, optimize for Bing's slightly different ranking signals (more weight to exact-match domains, on-page keyword density, and meta tags), and monitor Bing rankings alongside Google.

Direct-answer paragraphs (50 words, definitional, plain prose), question-formatted H2 headings with answer-formatted body text, FAQ sections with FAQPage schema, citation-friendly bullet blocks (3-5 short attributable claims per page), and original-data lines that name a specific number with a specific source. ChatGPT tends to skip walls of marketing copy and pull from clean, structured, attributable sentences. The same patterns work for Perplexity, Gemini, and Google AI Overviews — which is why this content layer compounds across all generative engines.

We use Mention Layer (Joel's SaaS) to monitor 30-50 keywords across ChatGPT, Perplexity, Gemini, Claude, and Google AI Overview weekly. For each keyword we track: was your brand mentioned, was it cited with a link, what was the citation context, what was the sentiment, and which page was the source. Then we benchmark against your top three competitors on the same keywords. The composite output is a citation share metric — your share of voice across AI answers in your category, tracked over time.

Both, in the right order. Google still drives the majority of click-through traffic for most B2B and local categories — don't abandon it. But Google's AI Overviews now appear above the organic results on ~13% of queries, and ChatGPT, Perplexity, and Claude are stealing the top-of-funnel research stage. The right framing: traditional SEO is the foundation, GEO and ChatGPT SEO are the additional layer that captures the buyer journeys SEO alone is no longer reaching. We run them together, not as a substitution.

ChatGPT Search and web-browsing citations can shift in 30-60 days because they pull from Bing's index, which updates quickly. ChatGPT's training-corpus citations move on longer cycles tied to OpenAI's model training schedule — we typically see meaningful lift on training-corpus mentions over 90-180 days, accelerated by tier-1 publication placements. The fastest wins come from Bing rankings + structural site upgrades. The compounding wins come from the third-party citations that train the model itself.