ChatGPT Ads Context Hints: What They Are, and What They Are Not
By The InPromptAds Team 5 min read
Last verified 15 August 2026. Version 1.0. Next scheduled review 15 September 2026.
Context hints are the closest thing ChatGPT ads has to targeting, and the reason most advertisers get them wrong is that the interface invites a keyword habit that the system does not reward.
OpenAI's own wording is the clearest statement available: hints help guide ad matching, but they are not exact-match keywords and do not guarantee delivery in specific conversations.
Read that as three separate claims, because each one kills a different assumption.
What OpenAI documents
Where they live. Advertisers provide context hints at the ad group level, describing the conversations, topics or keywords where their products or services may be relevant.
What the system considers when matching an ad:
- the context and intent of the current conversation
- the ad's landing page
- the ad's title and copy
- the advertiser-provided context hints
Your hints are one input of four, and the other three are things you also control but probably were not thinking of as targeting. Your landing page is a matching signal. Your ad copy is a matching signal. An advertiser who writes precise hints and points them at a vague landing page has undercut their own targeting with the input they spent the least time on.
The three assumptions that break
"Hints are keywords." They are not exact-match. Writing twenty single-word hints in the shape of a search campaign is applying a bidding model to a system that reads intent. Describing a situation is closer to the documented behaviour than listing terms.
"A hint guarantees I appear in that conversation." OpenAI states plainly that hints do not guarantee delivery in specific conversations. There is no equivalent of an exact-match term you can buy and expect to own.
"More hints means more reach." Nothing in the documentation supports this and the matching description argues against it. If the system weighs conversation intent against your landing page and copy, hints that pull away from what your page actually offers create incoherence between the four inputs rather than breadth.
Why we are writing this after saying not to
Worth being open about, since our own planning said otherwise.
Our Content Opportunity Map graded context hints as do not write on 1 August, on the grounds that the cluster was occupied by adsmurai, theadleaf and a purpose-built tool, and that adding anything required live account data we do not have.
That reasoning was about optimisation content, and it still holds. We cannot tell you which hints perform, because that needs an account and a spend history.
What changed is that OpenAI's documentation now states what hints are and what they do not do, and the corrective version of this page needs no account at all. The gap is not "how to write hints that win." It is that a large amount of published advice treats them as keywords, and the platform documentation says they are not.
What we cannot tell you
Which hints perform. Requires an account, campaign history and enough volume to compare. We have none of those.
How the four matching inputs are weighted. Not documented. Anyone stating a weighting is inferring.
Whether hint count affects delivery. Not documented, and we are not going to model it.
What "intent" means operationally. OpenAI names conversation context and intent as inputs. It does not define how either is derived, which is the part that would actually let an advertiser reason about matching.
Whether hints interact with the sensitive-context exclusions. No ads run beside conversations about personal health, mental health or politics, or in sensitive user journeys. Whether a hint pointing at an excluded topic simply fails or affects the ad group more broadly is undocumented.
Quick answers
What are context hints in ChatGPT ads? Descriptions of the conversations, topics or keywords where an advertiser's product may be relevant, set at the ad group level. OpenAI states they help guide ad matching but are not exact-match keywords and do not guarantee delivery in specific conversations.
Are context hints the same as keywords? No. OpenAI explicitly states they are not exact-match keywords. Matching also considers conversation context and intent, the ad's landing page, and the ad's title and copy.
How does ChatGPT decide which ad to show? Per OpenAI's documentation, matching considers the context and intent of the current conversation, the ad's landing page, the ad's title and copy, and the advertiser-provided context hints.
Can I guarantee my ad appears in a specific type of conversation? No. OpenAI states hints do not guarantee delivery in specific conversations.
Where do I set context hints? At the ad group level.
Does my landing page affect which conversations my ad appears in? Yes. OpenAI lists the ad's landing page as one of the inputs considered when matching an ad to a conversation.
How many context hints should I use? Not documented. OpenAI publishes no guidance on hint count, and any specific number circulating is derived from one advertiser's account rather than from the platform.
Sources
| Claim | Source | Tier |
|---|---|---|
| Hints guide ad matching, are not exact-match keywords, do not guarantee delivery in specific conversations | OpenAI Help Center, Ads in ChatGPT: The Basics, updated ~9 Aug 2026 | Confirmed, primary |
| Matching considers conversation context and intent, landing page, title and copy, and context hints | Same | Confirmed, primary |
| Context hints are set at ad group level | Same | Confirmed, primary |
| No ads beside personal health, mental health or political conversations, or sensitive user journeys | OpenAI ad policies v1.3, 15 July 2026 | Confirmed, primary |
| Hint performance, optimal hint count, input weighting | Not published | Absent |
Related reading
Context Hints Generator — draft hint candidates from a landing-page summary (not keywords).
How the ChatGPT ads auction works — campaign types, bids and budgets
Brand safety on ChatGPT ads — where ads can and cannot appear
Changelog
15 August 2026. v1.0 published. Reverses the Content Opportunity Map's do not write grading on this cluster, which applied to optimisation content rather than to the definitional correction.
Field kit
Tools
Site
OpenAI (primary)
The InPromptAds Team
Writes about AI-native advertising and the shift from search to conversation.
