Context hints add information at ad-group level about what a product offers, who it helps and when it may be useful. They are a contextual signal — not exact-match keywords, audience rules or a guarantee that an ad will appear in a specific conversation.
OpenAI’s documentation is explicit: context hints help the system understand needs, situations and natural language variations inside a focused ad group. They do not replace targeting, creative or the landing page; each layer has a different job.
It is additional information at ad-group level that helps describe what the product or service offers, who it helps and the situations in which it may be useful. OpenAI says it adds context beyond the headline, ad copy and landing destination.
Its role is to describe conversations, needs and themes the group is designed to support rather than trigger one exact matching rule.
Official source: Create Ad Groups for ChatGPT Ads · OpenAI Help Center.
OpenAI explicitly says context hints are not exact-match controls, audience-targeting rules or instructions to show an ad only in particular conversations. They do not guarantee delivery for specific words, topics, audiences or situations.
If you need real control over a first-party list, OpenAI now documents Custom Audiences as a separate mechanism for campaign-level inclusion or exclusion and ad-group bid adjustments. That does not turn a context hint into an audience.
| Element | What it does | What it does not do |
|---|---|---|
| Context hint | Adds context around product, need, situation and intent inside the ad group. | It is not exact match or an audience rule and it does not guarantee delivery. |
| Custom Audience | Uses first-party lists for inclusion, exclusion or bid adjustments. | It does not describe conversational intent by itself or replace context hints. |
| Ad copy | Communicates the proposition, benefit and reason to click to the user. | It is not a hidden targeting instruction. |
| Landing page | Continues the promise and contributes another relevance and conversion signal. | It does not define an exact matching rule. |
Official sources: Context hints and ad groups · Custom Audiences · OpenAI Help Center.
Delivery does not depend on the hint alone. OpenAI describes a broader relevance system that can use the current conversation’s context and intent, landing page, headline, copy, context hints, available controls and expected performance.
We therefore treat hints as one part of ad-group architecture rather than the only matching lever.
Official source: Ads in ChatGPT: The Basics · OpenAI Help Center.
OpenAI recommends keeping each group focused on a common product category, theme or customer need. Hints can cover related situations and natural variations within that idea.
If use cases would require materially different ads or landing pages, separate ad groups are clearer. OpenAI also recommends using multiple ads in each group to test different messaging approaches.
Official source: Ad-group best practices · OpenAI Help Center.
A useful hint adds information that is not already obvious from the ad or landing page, connects product features with real customer needs, describes situations rather than broad audience labels and uses clear natural phrases. It should also stay within genuine use cases for the product.
OpenAI’s drafting question is a good one: what else should the system know about this product, who it helps and when it is useful?
Official source: Writing effective context hints · OpenAI Help Center.
Useful specificity is not about adding more words. It is about clarifying who, which problem and which situation the offer genuinely fits, without pretending that the phrase becomes an exact query.
| Weak hint | More useful hint | What it adds |
|---|---|---|
| CRM | B2B sales team that needs to centralise opportunities and reduce manual follow-up. | User, problem and operating moment. |
| Accounting software | Small business with several bank accounts that needs to reconcile transactions and close the month with less manual work. | Context, friction and a concrete goal. |
| Industrial ERP | B2B manufacturer with distributed production and stock that needs to unify orders, availability and dispatch. | Real operational complexity and use case. |
Being specific does not mean the person must use those exact words. OpenAI explains that hints help the system understand broader needs and natural variations in language.
Official source: Context hints · OpenAI Help Center.
We use Search Console, paid-search queries, sales calls, CRM notes, support tickets, FAQs, comparisons and qualitative research to identify problems, constraints, goals and decision moments.
Keywords can provide useful language, but the work is to reconstruct complete situations the offer genuinely solves.
If the hint describes a specific need but the ad remains generic and the landing page tries to serve the entire market, the group loses coherence. The destination should quickly confirm why the offer fits that context.
The structure matters only if delivery and conversion can later be interpreted against a commercial hypothesis.
Ads Manager lets advertisers review performance in campaign, ad-group and ad views. The documentation does not frame each context hint as a keyword with its own exact query → result report.
We therefore evaluate the group as a hypothesis: delivery, CTR, CPC, conversion and downstream quality where volume is sufficient. A high-delivery group producing irrelevant traffic is not better than a smaller group producing commercial value.
Official source: Launch your first campaign · OpenAI Help Center.
There is no universal list of winning phrases. Product, market, context, creative and landing page all change the outcome. The compounding advantage comes from documenting which groups deliver, convert and create quality.
The useful loop is hypothesis → group → result → learning, not copying context-hint templates without context.
What they control, what they do not control and how to structure them for useful learning.
Additional ad-group information describing what a product offers, who it helps or when it may be useful.
No. OpenAI explicitly says they are not exact-match controls. They can contain relevant language, but their role is to describe broader needs and situations.
Not as an exact audience rule. OpenAI says context hints are not audience-targeting rules. If you need to include or exclude a first-party list, Custom Audiences is a separate campaign mechanism.
No. They help the system understand relevance but do not guarantee appearance for a particular word, topic, audience or situation.
OpenAI does not publish one universal number that is correct for every case. Coherence matters more: one group should not mix uses that require different ads or landing pages.
Not directly. Search terms can provide useful buyer language, but they should be developed into complete needs, situations and use cases.
When the product, use case, message or landing page changes materially. OpenAI recommends separating clearly different uses that require different communication or destinations.
By evaluating the ad group’s delivery, cost, conversion and downstream commercial quality. Ads Manager reports at campaign, ad-group and ad level; we do not treat each hint as an individual keyword with exact reporting.
We can audit ad-group structure, context hints, creative and landing pages so the account is built around hypotheses that can actually be measured.
What you sell, which markets you work in and what you want to learn from the first pilot. If the scope is not clear yet, we will define it with you.