ContextContext Hints FAQ for ChatGPT Ads
Direct answers first, followed by the focused guide, tool, or evidence page that resolves each question in depth.
Find the direct answer, then follow one route.
Each answer is intentionally short. The linked guide owns the complete explanation, examples, and current source notes.
A context hint is advertiser-provided descriptive text at the ad-group level. It describes conversations, needs, topics, products, or situations where the ads may be relevant and guides matching by meaning.
Describe one coherent buyer need or situation in natural language. Be specific enough to communicate real relevance, broad enough for wording variation, and include only constraints that materially change product fit.
Current public OpenAI guidance does not publish a universal ideal length in the sources reviewed on July 22, 2026. Use the shortest wording that preserves the real need, situation, and fit constraints.
OpenAI does not prescribe one universal count. Keep related variations within one focused ad-group theme and separate meaningfully different products, audiences, use cases, or intent stages.
No. OpenAI describes context hints as broad thematic signals rather than exact-match rules. They describe relevant meaning; they should not be treated as a keyword list or assumed to provide exact exclusions.
OpenAI's current checklist represents them as a JSON list. During a bulk update, provide the complete intended list because the uploaded value replaces rather than merges with the existing list.
Low delivery is not automatically a hint-writing problem. Check account and campaign status, approval, dates, geography, bid, budget, policy eligibility, creative, landing page, and available opportunity before rewriting.
Compare separate, interpretable ad-group hypotheses and evaluate qualified business outcomes after confirming comparable eligibility and opportunity. Public captured-ad evidence cannot reveal private conversion performance.
No. They are evidence-backed inferences reconstructed from captured ads and associated prompts. They are labeled as inferred and should not be presented as literal advertiser-submitted settings.
No. ContextHint generates and researches targeting hypotheses. Its MCP supports repeatable research inside an AI client but does not access Ads Manager, publish campaigns, change bids, or edit live ad groups.