A fictional outdoor-equipment business has two customer situations: carrying supplies to a picnic and moving heavy tools around a worksite. Both involve transporting objects. That shared verb does not establish that the same offer or context hints should describe both situations.
The useful boundary is the relationship between the need and the advertised offer. A lightweight folding cart may fit one situation and not the other. If both needs rely on different load requirements or different products, a broad “transporting equipment” brief can conceal the difference.
OpenAI documents context hints at ad-group level in its campaign management guidance. This article focuses on the editorial decision about which needs form a coherent brief. It does not replace the wider decisions about account structure, bidding or targeting.
Build cards before building groups
Give each candidate need a short card containing the task, relevant constraint and actual offer that supports it. Avoid starting with broad category names such as outdoors, productivity or business services. They can collect unrelated tasks without explaining the connection. If it is still unclear whether the descriptions represent different decisions, map the boundaries between the customer needs before grouping the brief.
For the fictional cart business, one card might describe carrying picnic supplies across a park, with folding storage as the relevant constraint. Another might describe moving worksite tools, with a documented load requirement. Check both against the real product facts before considering a shared brief.
The card should also name the advertised product or package. “Our range can do this” is not specific enough if the group contains ads for only part of that range. The editorial grouping needs to match what the customer will actually be offered.
Transport is not the whole need
- Picnic
Being able to fold the cart may matter between uses.
- Worksite
Load and working conditions may impose different requirements.
- Group boundary
A shared verb is insufficient when the offers need to differ.
Checkpoint
Ask whether one explanation remains true
As an editorial check, try writing one sentence that accurately explains why the offer fits all the cards in a proposed group. If the sentence remains concrete and meaningful, the grouping may be coherent. If it becomes a vague category label, you may be combining needs that deserve separate treatment.
For example, “portable seating for temporary indoor meeting setups” might cover several meeting-room situations if the advertised chairs support them. Adding outdoor festival use may introduce weather exposure, ground conditions and different use requirements. The common word “seating” does not resolve those differences.
Do not force every minor variation into a separate group either. Two customer descriptions can use different words for the same supported task. “Pack away seating after a meeting” and “clear a room after a session” may belong together when their requirements and offer are the same.
Use a counterexample to test the boundary
For each proposed group, invent a plausible case that sits just outside it. This is a reasoning exercise, not a claim about market demand. Ask whether the current description accidentally includes that case.
| Proposed shared need | Counterexample to examine | Why it matters |
|---|---|---|
| Temporary indoor seating | Seating stored outdoors between events | Storage conditions may differ |
| Reviewing a draft with external clients | Publicly publishing the final document | Review and publication are different tasks |
| Short equipment hire for one event | Retaining equipment for an uncertain project | Duration and extension conditions may differ |
If the counterexample appears to fit the words but not the offer, sharpen the boundary. Add the relevant constraint or separate the candidate. This is more informative than choosing a group name that merely sounds tidy.
Check shared context against each creative
Lay out the proposed hints alongside the ads that would share them. For each ad, ask which candidate needs it honestly addresses. An empty or uncertain answer identifies a possible mismatch.
This review can reveal two different problems. The hint list may include a need unrelated to the advertised offer. Or the group may contain an ad for another package whose requirements differ. Resolve the actual problem rather than deleting useful detail until everything appears broadly compatible.
Use the hint-to-promise review for the individual comparison. The grouping decision asks a wider question: can the set be described coherently without misleading any of its members?
This method detects contradictions in the brief; it is not a documented rule requiring every hint to repeat every ad. Two ads can emphasise different uses of the same offer. The editorial question is whether the list describes the actual offers without borrowing a capability from another package. OpenAI’s targeting documentation describes hints as additional context. It does not guarantee how a particular phrase will distribute delivery between the ads.
Keep a boundary note, not just a name
Record what the group includes and why. Also record the nearest excluded situation and the reason it sits outside. “Includes client review of drafts; excludes final public publishing because that is a separate workflow” is more useful than a name such as “Collaboration”.
The excluded case belongs in the internal boundary note. It is not a negative-keyword rule to submit to the platform. This note helps when a new creative arrives. The team can compare the offer and task with the existing boundary rather than adding it because the product category looks related. Revisit the boundary when the actual offer changes, not only when the list becomes long.
Once the grouping makes sense, use semantic deduplication to remove repeated descriptions within it. Deduplication is a later task: two rows can mean the same thing even inside a well-defined group, while a clean list can still combine incompatible needs.
If the cards are too vague to judge, return to product-to-use-case translation. A coherent hint brief starts with specific supported relationships. Its quality is not determined by having the fewest groups or the most detailed labels, but by whether another reviewer can explain why each included need belongs with the actual ads.
Sources and scope
Deciding whether several customer needs can share one coherent context-hint brief.
Working methods and examples are editorial suggestions. Check current platform requirements and available features before implementation.