The builder starts from five core intent buckets that apply to every healthcare specialty on ChatGPT Ads: awareness/symptom-stage queries, high-intent procedure queries, comparison and consideration queries, pricing and package queries, and location/proximity queries. These map to the five stages a real patient conversation moves through inside ChatGPT before they book — from "why does my tooth hurt" to "book me an appointment near me."

Your specialty selection swaps in real procedure and unit terminology — a dental clinic gets "root canal" and "implant" example queries, an IVF centre gets "IVF cycle" and "egg freezing," and so on — so the example queries in each bucket read like something a real patient would actually type or say inside a ChatGPT conversation, not generic keyword-tool output.

Each buyer intent you select — second-opinion seekers, NRI patients, insurance-covered treatment, emergency care, follow-up and recall — adds one dedicated extra campaign bucket on top of the five core buckets, capped at eight total so the structure stays manageable for a single account manager to run. Budget is then allocated across all buckets by intent value: high-intent procedure queries and pricing queries get the largest share, since they convert closest to a booked appointment, while awareness and comparison queries get smaller test budgets to build conversation-completion signal without overspending on browsers.

The output is a starting structure, not a finished media plan — real accounts get refined weekly against conversation-completion attribution data once they are live. But it gives you a defensible, intent-separated skeleton on day one instead of one flat campaign competing against itself.