How to Get Your Brand Recommended by ChatGPT

Two layers produce the answer. Training: what the model absorbed, which is why the content licensing deals matter (Google licensed Reddit's data in February 2024, OpenAI followed in May 2024). Retrieval: the live search the assistant runs before answering, with cited sources. You influence training slowly and retrieval faster, and both point at the same conclusion: recommendations trace to third-party sources the assistant considers credible. Profound's analysis of 4+ billion citations finds Reddit the most-cited domain of all.

Your own site claiming you're the best is exactly the evidence assistants discount. What moves recommendations is strangers saying it in citable places: community threads, comparison content, reviews. The profile of what gets cited is documented and counterintuitive: Semrush's study of 248,000 AI-cited Reddit posts found 80% had fewer than 20 upvotes, with the average cited thread roughly 900 days old. Small, specific, exactly-on-topic beats viral. Earning those mentions honestly is slow, which is why it's defensible; the mechanics are in our Reddit SEO guide and it's the engine under our AI visibility service.

Assistants do read your site, mostly for facts: what you cost, who you're for, what's included. Make those facts extractable: answer-shaped pages where one clear claim sits under one clear heading, FAQ blocks that say the thing plainly, schema markup, and an llms.txt file stating the offer in plain text. Citable structure doesn't win the recommendation by itself, but it decides how accurately you're described when you are mentioned, and wrong price quotes from stale pages are a self-inflicted wound.

Boring but load-bearing: the same name, the same one-line description, the same category everywhere the machine looks (your site, profiles, directories, the threads you answer in). Assistants cross-reference; contradictions dilute the entity and can split you into two half-known brands. Before building new mentions, make the existing ones agree with each other.

Buying placement (nobody sells it honestly; there is no AdWords for ChatGPT's organic answer). Stuffing "best X" claims on your own pages (self-claims are the discounted evidence). Prompt-injection tricks hidden in page text (patched, and radioactive to your brand when found). Astroturfed threads and fake reviews (Reddit's rules ban the mechanics, and the ban converts your budget into a reputation problem). And over-fitting to today's weightings: Semrush watched ChatGPT's Reddit citation rate swing from ~60% to ~10% of responses inside two months. Chase the inputs, not the current tuning.

A fixed panel of buyer questions, run monthly across ChatGPT, Perplexity and Google's AI surfaces, recording verbatim answers and cited sources, with competitors scored on the same panel. That's the whole method, and you can run it yourself in an hour: the procedure is in our DIY audit guide. Expect movement in quarters, not weeks; mentions have to be created, then indexed, then retrieved.