Hints in ChatGPT Ads: from buying keywords to understanding intentions

For years, digital advertising has operated under a linear and familiar logic: a user searches for a term, a brand bids on that keyword, and the platform displays an ad. This model, although imperfect, has supported agencies, brands, and search networks for decades.

However, conversational environments completely disrupt this dynamic. With the arrival of ChatGPT Ads, OpenAI introduces a native model where ads appear while the user explores, compares alternatives, or matures a decision within a chat.

OpenAI’s policy is clear: Ads are displayed separately from responses, transparently labelled, and do not influence the organic content generated by AI. Advertisers cannot modify, prioritise, or alter the model’s responses.

Under this new ecosystem, the question for brands changes completely:

• Before: What keywords do I want to buy?
• Now: In which conversations and contexts is my brand genuinely useful?

This is where Hints come into play.

What is a “Hint” in OpenAI Ads?

A hint is a contextual signal that guides the advertising matching system to identify conversations where an ad may be relevant.

Unlike a traditional keyword, OpenAI uses these context hints within ad group settings to map the relevance of a product or service against the flow of a live conversation. They do not work through exact matches of isolated terms, such as “CRM” or “trainers”; instead, they describe a situation of need.

For example, saying “reporting software” is not the same as saying “marketing teams that need to centralise campaign data and reduce the time spent on manual reporting”.

The first option is a keyword. The second describes a context, an intention, and a problem.

And in conversational advertising, that matters much more.

Comparison table: Keyword vs. Conversational Hint

The paradigm shift: From isolated words to real situations

In a traditional search engine, the user summarises their need in short commands (best reporting software). In ChatGPT, intent is expressed in a more natural and narrative way:

"My sales team spends too much time manually updating reports in Excel, and we need a simpler way to automate month-end processes."

The user has not written the word “buy”, but they have described the exact problem your software or agency solves. Generative systems do not interpret isolated terms; they interpret relationships: who is asking, what problem they are describing, their level of frustration, and the type of solution they expect.

Mapping real conversations and intentions

• User conversation: "I don’t know which campaigns are actually driving sales and which ones are just noise."
• Underlying intent: Measurement, attribution, incrementality, MMM models.
• User conversation: "It takes us weeks to adapt creatives for each channel and market."
• Underlying intent: Creative automation, asset scaling, AI design tools.
• User conversation: "I want to know whether my brand is appearing in ChatGPT responses."
• Underlying intent: Brand tracking, AI visibility, generative Share of Voice.

A practical framework for structuring your Hints

To avoid treating hints as simple long-tail keywords, use the following context architecture formula:

(HINT) = USER + NEED + CONTEXT + INTENT MOMENT

Example of the framework in action:

• User: Digital marketing teams.
• Need: Automate reporting.
• Context: High-volume multichannel campaigns.
• Intent moment: Comparing tools in the market.
• Resulting hint: "Digital marketing teams that are evaluating and comparing tools to automate multichannel campaign reporting and reduce manual processes."

Ad group strategy based on intent stages

Not all conversations have the same commercial value or happen at the same moment. Hints should be segmented into differentiated ad groups based on user maturity:

1. Aspiration stage (Discovery)

The user explores possibilities, trends, or general improvements.

• Hint example: "Users looking for ideas or methodologies to improve the productivity of a paid media team using AI."
• Ad focus: Educational content, thought leadership, value-driven guides.

2. Consideration stage (Evaluation)

The user identifies the type of solution they need and compares direct alternatives.

• Hint example: "Brands comparing solutions or agencies to measure their visibility in generative AI responses."
• Ad focus: Technical differentiation, sector-specific success stories, competitive advantages.

3. Decision stage (Conversion)

The user is ready to take a commercial or contracting action.

• Hint example: "E-commerce leaders interested in requesting a demo or immediately implementing catalogue feed automation."
• Ad focus: Reduced friction, direct value proposition, demos, agile contact.

Best practices for campaign setup

• Use natural language: Write hints the way your ideal customer speaks, not as an internal taxonomy from your database.
• Add high-intent verbs: Words such as compare, evaluate, implement, integrate, replace, or request help the algorithm distinguish purely informational traffic from transactional traffic.
• Do not limit context to cost: Avoid competing only on price within the hint text. AI looks for situational affinity; if you provide infrastructure context, such as "connect with Salesforce", activation will be much more precise.

The impact on marketing and measurement teams

The adoption of OpenAI Ads and its hint system transforms the internal operations of teams:

1. Breaking down silos (Paid Media + SEO + Content): The paid media specialist can no longer work in isolation. They directly depend on content and strategy teams understanding the real problems, objections, and questions of the audience in order to translate them into precise hints.
2. A closed measurement ecosystem: OpenAI already offers documentation and advanced measurement tools for advertisers, including its own conversion pixel and conversions API. This makes it possible to connect visibility within conversations with real return on investment (ROI) in the business backend.
3. The need for external authority: In an environment where the user is actively interacting, the ad must provide real value. Interruption is penalised by user rejection. If the ad does not align with the solution to the problem raised in the chat, performance can drop significantly.

Conclusion

Hints represent a shift in mindset for advertising activation. We are moving from a keyword-based logic to one based on conversational intent.

For brands, this means improving their content, structuring their data more effectively, understanding their audiences better, and measuring their presence in new decision-making environments.

In OpenAI Ads, value will not only come from appearing when someone searches for a solution, but from being present when someone is describing the problem that solution can solve.

And this is where brands capable of connecting context, technology, creativity, data, and measurement will gain a clear advantage. Those that continue thinking only in terms of keywords can still call it a “legacy strategy” and make it look good on a slide.