ChatGPT is OpenAI's flagship — roughly 400 million weekly users, the single largest place on the internet where consumers ask "who should I call." When the question is "best personal injury lawyer in [city]," GPT-4o returns three to five firms. Here's how it decides.
ChatGPT does not perform a real-time search of the web for every query. It blends three sources: pre-training data (the snapshot of the web the model learned from), retrieved context (real-time browsing for queries that warrant it), and reinforcement learning patterns. For "find me a lawyer" queries, ChatGPT increasingly uses retrieval — meaning it fetches current data — but the model still draws heavily on training-data familiarity to decide which firms are credible enough to surface.
The implication: a firm that was rarely mentioned during the training window is less likely to be named even if it has invested heavily in citations since. New citations help — the retrieval window picks them up — but the entity recognition that comes from being widely cited during training is harder to retroactively build.
ChatGPT is queried with a wide vocabulary, but for consumer "find a PI lawyer" intent, a small handful of prompt patterns generate the majority of decision-stage queries. The diagnostic tests these against your firm:
The firms ChatGPT names most consistently across these prompts are not always the largest firms in a metro. They are the firms with the densest, cleanest entity surface across the legal web — a pattern that has very little to do with traditional marketing spend.
The work is the standard AEO stack with three ChatGPT-specific emphases. First, build citation density on the authority sources the model trusts most heavily — Justia, Avvo, Super Lawyers, your state bar's public directory, and any practice-area vertical publications. Second, resolve every entity ambiguity. Pick the canonical firm name and unify all citations to match. Third, build narrow practice-area-by-jurisdiction pages and submit them to retrieval-friendly sources.
Velocity: a firm starting at zero ChatGPT mentions can typically achieve named appearances in 8-16 weeks with focused work. The signal that the firm is "trusted enough to name" lags the actual citation activity by roughly a retrieval cycle.
The diagnostic tests the consumer prompts that matter — and shows you the answer.
Run the Diagnostic →