Skip to content

SWLR Articles

Industry News: Reports reveal ChatGPT ads that open branded conversations

Brands may need helpful conversational product guidance and new success metrics when an ad interaction stays inside the assistant.

Rob Kerry

TLDR;

ChatGPT is testing ads that start a chat with a brand's AI agent, according to Digiday. Customers stay in ChatGPT instead of visiting the brand's website. It's an early test with selected clients, so brands should focus on useful answers and progress towards a purchase, rather than clicks alone.

What happened

Digiday reported a ChatGPT ad format shown to selected clients in which a click starts a conversation with a brand’s AI agent inside ChatGPT. The report described an early iteration, not confirmed broad availability; this date is the report date because the initial introduction date was not established.

Why it matters

The reported format would keep an initial paid interaction inside ChatGPT, reducing the information a conventional landing-page funnel can observe. The commercial opportunity depends on whether the brand's agent provides accurate guidance and a qualified next action. Digiday's report describes an early limited-client iteration, so it should not be presented as generally available inventory.

A click that starts a brand conversation changes what the customer receives from the ad. Instead of arriving on a page with a fixed explanation and visible calls to action, the user enters a service that must respond appropriately to questions. That can reduce effort when the agent helps with a bounded task, but it can also keep the interaction away from the website analytics used to judge conventional campaigns. The enterprise needs a clearer definition of progress: an accurate answer, a suitable product choice or a qualified handoff, rather than another message in the conversation. Digiday's reporting describes an early selected-client iteration, so implementation plans should remain conditional on verified access and product requirements. The interesting commercial question is whether the conversation helps the customer achieve something useful enough to justify the media cost.

How your brand can benefit / be affected

Verify current participation and product requirements before proposing implementation. Define a bounded customer task, reliable product evidence, permitted data use and an appropriate escalation or handoff.

Evaluate task completion, guidance accuracy and qualified outcomes rather than clicks alone. Clarify what aggregate or event reporting the platform actually supplies. Keep the report date as the public news date and distinguish confirmed capabilities from assumptions about future conversational ad formats.

Choose a task the brand can support with reliable evidence. A product-guidance agent might clarify a compatibility requirement or distinguish two suitable configurations; it should also say when the offering does not meet the customer's need. Define approved facts, boundaries and the point where a person or another channel should take over. Check what permissions and reporting the actual programme supports before designing data collection or a measurement funnel. Do not assume the brand receives the whole ChatGPT conversation. Treat the agent as a customer-facing service with named owners for knowledge updates, errors and escalation, rather than a more interactive piece of ad creative.

Set success measures that reflect the chosen task and the evidence available. Review guidance accuracy, correctly completed tasks and qualified next actions, then connect downstream outcomes where the supported handoff permits it. Message volume or conversation length can indicate engagement without establishing value. Include the cost of maintaining and supporting the agent when judging campaign economics. Keep ordinary website clicks as one possible outcome rather than the only definition of success. Verify participation before committing implementation resources, and label unconfirmed capabilities as assumptions in the business case. The brand should proceed when the available format can deliver a useful conversation and an accountable commercial outcome, not merely because it keeps the user inside ChatGPT.

News date: 14 September 2026. Editorial review: 16 September 2026. Analysis includes subsequent developments where stated.