Contextual matching

Google uses the term for ads in AI Overviews and AI Mode: the system can match an ad to the query plus the AI-generated response, alongside direct query matching. AI Max for Search, broad match, Shopping and Performance Max campaigns are eligible for both. Exact and phrase match keywords are not eligible for contextual matching on exploratory questions.

Gravity matches every ad on context. Its engine compares recent conversation messages with active campaigns, runs an auction and returns the winning ad. Gravity’s docs say matching works on context rather than keywords, and its campaign API rejects a keywords field, although its advertisers page mentions keyword targeting; the sources disagree.

ChatGPT ads take a related approach through context hints, which describe relevant products, use cases or needs and are explicitly not exact-match keywords. OpenAI states that conversations are not shared with advertisers, who receive aggregated, non-identifying performance data only.

Contextual matching rewards descriptive inputs. Describe the problem your product solves in plain language, keep negatives and exclusions up to date, and expect search terms and contexts you did not list. Since you never see the conversations themselves, judge matching by aggregated outcomes such as CTR and conversions by ad group. In Admux, one campaign becomes one leg per channel, and each leg is matched by that channel’s own system, so the same brief reaches Google, ChatGPT and Gravity through three different matching methods. GET /channels shows which targeting inputs each channel accepts through Admux.