ChatGPT Ads targeting is built around conversational relevance. Advertisers use ad groups and context hints to describe the kinds of conversations where an ad should be eligible. This is closer to intent mapping than keyword bidding, so the best structure starts with user problems, comparisons, and decision tasks.
Audience problemConversation typeContext hintAd copyLanding pageConversion goal

What are context hints?

Context hints are advertiser-provided descriptions of relevant conversation types for an ad group. They help define when an ad may be relevant, but they should not be treated as exact-match keywords or a substitute for useful landing pages.

How is this different from keyword targeting?

Keyword targeting matches explicit search terms. Context hints describe the surrounding situation, problem, or task inside a conversation. A user may not say "hire a tax lawyer," but may describe a tax problem that makes a legal service relevant.

How should advertisers create intent maps?

Advertisers should group likely conversations by user job: learn, compare, diagnose, plan, buy, or troubleshoot. Each group should have a matching ad, landing page, and conversion goal.

What makes a bad context hint?

A bad context hint is too broad, too keyword-stuffed, or disconnected from the landing page. "People interested in marketing" is weak. "A Shopify owner comparing ways to improve paid search tracking before scaling ad spend" is stronger.

How many ad groups should a pilot use?

A first pilot should usually use a small set of clearly different ad groups. Too many ad groups makes early data hard to read. Too few ad groups hides which conversation types are actually working.

Context hint examples by intent

Intent typeWeak context hintStronger context hint
LearnUser is interested in softwareUser is learning how to choose project management software for a remote team
CompareUser is comparing productsUser is comparing two ecommerce platforms by migration effort, cost, and SEO risk
DiagnoseUser has a problemUser is trying to understand why paid search traffic is not converting
PlanUser wants helpUser is planning a 90-day paid media test for a new acquisition channel
BuyUser wants productsUser is evaluating product options and needs pricing, specs, and shipping confidence

Context hint QA checklist

  • Map the user job before writing hints
  • Use specific nouns and scenarios
  • Avoid keyword stuffing
  • Match each hint to a landing page
  • Separate learn, compare, diagnose, and buy intents
  • Keep first test structure readable
  • Review performance by ad group weekly