OpenAI ads need intent data to compete for search budgets

By Tom Whitfield

Google generated nearly $295 billion in advertising revenue last year. Search and other properties accounted for $225 billion, or about $4.3 billion a week.

That weekly figure exceeds the $2.9 billion The Trade Desk booked for a full year. Patrick Briggs, CEO of Semify, identifies The Trade Desk as the largest independent demand-side platform and expects OpenAI to reject the Google comparison publicly. His case rests on search advertising economics: OpenAI wants the budgets Google already captures.

What has OpenAI already launched?

OpenAI has started selling advertising through a beta, self-service Ads Manager. Advertisers can buy on a CPM or cost-per-click basis, measure conversions with a pixel and Conversions API, and view campaign-level performance reporting.

OpenAI also tests Sponsored Agents. A user who clicks an ad can begin a separate conversation with the advertiser, extending the interaction beyond the placement.

OpenAI says advertisers receive aggregated performance insights without access to individual chats. That addresses a basic privacy condition, but it does not yet show advertisers what prompted an impression or whether the exchange represented buying intent.

Why ChatGPT prompts matter

Briggs contrasts passive browsing with expressed demand. People open Instagram to scroll; they open Google because they need a blue sweater. Many use ChatGPT for specific searches, not loose browsing.

A prompt may describe something warm for a week in Chicago in January, reject wool because it causes itching and set a $150 spending limit. Those details form a brief built from preferences, constraints, objections and purchase criteria.

That information gives OpenAI what Google spent years teaching people to provide. Briggs argues that OpenAI has solved the consumer side of search advertising because users reveal goals in their own words. His words: "We need to know which kinds of needs, at which decision stage, are producing results."

What does this change for media buyers?

Nothing yet supports treating ChatGPT as a Google Search replacement. Briggs reports that advertisers see an ad beside a conversation without knowing the conversation type. Google provides search-trigger reports, customer conversions and negative keywords to block wasteful terms.

Briggs wants ChatGPT reporting to show whether impressions reached people exploring, comparing or ready to act, and which product criteria drove engagement. He also wants advertisers to see which sweater prompts produced business and which did not. Google advertisers have had some version of that map for two decades, he says.

Budget allocation follows evidence. A chief marketing officer can file ChatGPT under experimental or awareness spending when the platform offers reach, impressions and clicks without meaningful intent data. That money comes from innovation budgets or paid social. Connect privacy-safe intent categories with business outcomes, and ChatGPT can compete for paid search dollars set aside to capture existing demand.

Can privacy limits still support intent reporting?

Yes, according to Briggs and established search practices. People may tell ChatGPT details they would never type into a search box, and no advertiser should read individual conversations or prompts. Google limits, aggregates and buckets queries too rare to display, while advertisers still receive useful demand signals.

Briggs proposes extending that model to conversations. Advertisers would see aggregated commercial patterns organized by intent category, while individual chats remain locked. The ad industry already produces pseudonymized, aggregated audience and intent signals, so he sees no need to expose personal data.

What should advertisers test now?

Briggs advises running ChatGPT against paid search and paid social to see which channel it resembles. He asks marketers to question their representatives about the intent reporting available now, then keep asking as the product develops. A click proves curiosity, but it does not justify search-level pricing without visibility into the intent behind it.

For now, media buyers should classify ChatGPT advertising as experimental or awareness work rather than a paid-search replacement. Better measurement could also improve the experience for users: someone seeking an itch-free sweater below $150 could receive a more relevant answer.

OpenAI still needs an advertiser-facing map of intent before it can claim Google's economics. Briggs set the condition: "It has to make that intent measurable without making it personal."