The digital advertising industry built its measurement infrastructure on a single interaction: the click that moves a user from an ad to a marketer's controlled environment. Every downstream conversion metric — session time, cart adds, purchases, email capture — assumes that click succeeded in transporting the user off the publishing surface. OpenAI's new Sponsored Agent format, described by Digiday on 14 September, is designed to make that click unnecessary.

The format works as follows: an ad unit inside ChatGPT carries a call to action — "Chat with us," "Find your product," "Get a recommendation" — that does not open a browser tab or redirect to a landing page. It opens a branded conversation window inside the ChatGPT interface, connecting the user directly to the brand's AI agent, which can answer product questions, handle returns, provide sizing guidance, or assist with purchase completion without the user leaving the platform. Wayfair is among the earliest testers, conducting limited-scale experiments with guardrails around product accuracy and service handoffs. OpenAI CFO Sarah Friar described the format as "truly endemic" AI advertising — positioning it as native to the AI surface rather than adapted from web display conventions.

The structural implication for performance marketing is significant. The click-to-site model produced a clean attribution chain: ad impression → click → landing page session → conversion event → revenue. Every link in that chain was instrumented by the marketer. With Sponsored Agent, the first decision point — whether the user's needs are met — occurs inside OpenAI's platform, under OpenAI's data architecture, using the brand's AI agent but not the brand's analytics stack. The user may purchase, may close the window, may return to ChatGPT, or may open a browser tab later with no tracked referral. First-party data collection — the email capture, the session identifier, the cart item — requires the user to have arrived on a marketer-controlled surface. In-chat engagement eliminates that arrival.

For brands approaching the winter holiday season — ten weeks from the date of this issue — the timing is acute. The Q4 planning cycle requires ad format decisions now: budget allocation, creative briefing, measurement frameworks, and performance targets are set in September for execution beginning in October. Sponsored Agent requires fundamentally different success metrics from any format currently in widespread use. Conversation engagement rate, AI agent resolution rate, and in-chat purchase completion are the relevant KPIs; click-through rate and session time are not applicable. Marketers who have not established baseline measurement infrastructure for conversational formats before Q4 campaign launch will be unable to optimise them during the peak spending window.

ROI measurement for ChatGPT advertising formats more generally remains unsettled. Existing ChatGPT ad products have not demonstrated consistent returns across verticals, and the measurement frameworks for comparing in-chat engagement to click-to-site conversion are immature. Wayfair's participation suggests a category — product discovery with high information density around dimensions, configuration, and delivery — where in-chat assistance is credible. The return on that engagement relative to the same spend in Google Shopping or Meta conversion campaigns is unknown.

The broader trajectory the format signals is unambiguous. The value proposition of AI advertising, from OpenAI's commercial perspective, is that the platform owns the user's decision-making environment at the moment of intent formation. A user asking ChatGPT "what sofa should I buy for a small living room" is expressing purchase intent in a context where a brand's AI agent can address the specific query, handle objections, and complete the transaction without the friction of a page load, a form, or a checkout flow. If the format can demonstrate conversion rates that justify its premium over standard display, it will redirect a meaningful share of Q4 performance budgets from Google and Meta to OpenAI. If it cannot, it will represent a format-development experiment that brands finance.

For GEO practitioners specifically: the Sponsored Agent format creates a commercial incentive for OpenAI to surface product content it can monetise over product content it cannot. A brand with a Sponsored Agent arrangement has a different relationship to ChatGPT's product recommendations than a brand without one. Whether that differential propagates into organic AI responses as well as paid formats is the structural question that determines whether GEO remains a merit-based discipline or becomes a pay-to-play surface. The answer will emerge from how Sponsored Agent performs in Q4 — and whether the format scales.