Short answer: ChatGPT Ads targeting is based on conversational context, campaign settings, ad group themes, landing pages, ad titles, ad copy, and context hints. Context hints describe the conversations, topics, needs, or situations where your product may be relevant — but they are not exact-match keywords and do not guarantee delivery in specific conversations.
What are context hints?
Context hints are broad thematic signals inside ChatGPT Ads ad groups. They help explain the types of conversations where an ad group may be relevant. A good context hint describes user need, product use case, pain point, or decision scenario.
| Weak hint | Better hint | Why it is better |
|---|---|---|
| Google Ads | Businesses trying to understand why Google Ads spend is rising without more leads | Describes the user's situation |
| Protein powder | Shoppers comparing high-protein powders by cost per serving and ingredients | Describes comparison intent |
| Golf trips | Golfers planning a multi-day trip and comparing destinations, course prices, and tee times | Captures task and decision need |
| Water testing | Homeowners trying to understand lead, PFAS, or nitrate risks in local drinking water | Captures specific anxiety and topic |
How are context hints different from keywords?
| Attribute | Search keywords | ChatGPT context hints |
|---|---|---|
| Format | Short terms or phrases | Descriptive conversation themes |
| Matching | Query-based | Conversation and intent-based |
| Control level | More explicit | Less exact, more semantic |
| Best use | Capture known demand | Match user situations and needs |
| Example | "google ads audit" | "Businesses looking for help diagnosing wasted Google Ads spend, tracking issues, and poor lead quality" |
How should you group ad groups?
Group ad groups by meaningfully different user intent. Do not combine unrelated products, audiences, or use cases in the same ad group. OpenAI recommends keeping each ad group focused on a single product category, theme, or intent area.
| Situation | Split or combine? | Reason |
|---|---|---|
| Same product, same buyer, same use case | Combine | Creative variations can test angles |
| Same product, different use cases | Split | Context hints and landing pages differ |
| Different products in same category | Usually split | Product relevance matters |
| Lead-gen and ecommerce offer | Split | Conversion paths differ |
| Awareness and direct response | Split | Objective and reporting differ |
Example: Google Ads consultant ad groups
| Ad group | Conversation theme | Landing page |
|---|---|---|
| Wasted spend audit | "I'm spending more on Google Ads but leads are not improving" | Google Ads audit page |
| Tracking problems | "GA4 and Google Ads conversions do not match" | Tracking audit article or service page |
| PMax cannibalization | "Performance Max is taking credit for brand sales" | PMax audit page |
| Consultant vs agency | "Should I hire a Google Ads consultant or agency?" | Consultant comparison page |
| Microsoft Ads expansion | "Should I import Google Ads into Microsoft Ads?" | Microsoft Ads comparison page |
Example: ecommerce travel bag ad groups
| Ad group | Context hint angle | Better destination |
|---|---|---|
| Carry-on bags | Users comparing carry-on bags for short trips and airline limits | Carry-on collection |
| Business travel | Users packing for work trips and needing laptop compartments | Business travel collection |
| Gifts | Users looking for practical gifts for travelers | Gift guide |
| Lightweight travel | Users trying to pack lighter for weekend trips | Lightweight bags collection |
| Premium luggage | Users comparing durable premium travel bags | Premium collection or comparison page |
Context hint formula
Use this structure to write context hints that give the system useful signals:
[Audience] + [problem or task] + [decision context] + [product or service relevance]
Example: "Ecommerce marketers trying to understand whether Performance Max is taking credit for branded search, remarketing, or existing shopping demand."
Example: "Ecommerce marketers trying to understand whether Performance Max is taking credit for branded search, remarketing, or existing shopping demand."
Common mistakes
- –Writing one-word context hints.
- –Copying Google Ads keyword lists into context hints.
- –Combining unrelated products in one ad group.
- –Using context hints that do not match the ad copy.
- –Sending all ad groups to the homepage.
- –Failing to create different creative for different user needs.
Related ChatGPT Ads guides
- → How to run ChatGPT Ads — setup, campaign structure, and launch checklist
- → ChatGPT Ads examples — ad copy, context hints, and landing page templates
- → ChatGPT Ads not delivering — troubleshooting checklist for low impressions and delivery issues
- → Brand search cannibalization in PMax — understanding intent matching in Google's AI campaigns