Ad tech does not wait for a platform to finish building before it shows up. Weeks after OpenAI began telling European ChatGPT users that ads were coming to the Free and Go tiers, and days after it confirmed the first market to go live was the United Kingdom, at least one established marketing platform had already run a pilot through the channel in the United States. AdRoll was not waiting to see how the room turned out. It was already testing furniture.
The pilot behind the headline
On August 11, AdRoll announced a pilot giving a select group of its small business, mid-market and B2B advertisers early access to buy placements inside ChatGPT. The company framed it as an extension of its existing multichannel platform rather than a standalone experiment, which is the more telling detail. AdRoll is not building a new ChatGPT-specific product. It is plugging an existing demand-side tool into a new distribution surface, the same move it has already made for connected TV and digital out-of-home.
“Conversational AI platforms are creating an entirely new environment that enable marketers to participate in the moments when options are being discovered, preferences are being formed, and decisions are taking shape,” said Vibhor Kapoor, chief executive officer of AdRoll. “Marketers need to understand the impact conversational AI can have on the middle of the customer journey and how it complements the channels they already use to build awareness and capture demand.”
Hans Fischmann, AdRoll’s vice president of product, put the internal read on early results more bluntly: “It is not every day that a new channel emerges, let alone one that’s tied to how people actively explore and evaluate their options through AI. The early signal is encouraging, and we are seeing ChatGPT show up across both awareness and performance goals.”
Built by the vendors who already built retargeting
AdRoll is not the first ad tech company to get a seat inside ChatGPT, and that is the actual story. Criteo became OpenAI’s first advertising technology partner back in March. AppsFlyer and Adjust, the two dominant mobile measurement platforms, are already wired in to give advertisers attribution data on a channel that otherwise looks like a black box. None of these are startups chasing a novelty integration. They are the incumbents that already run programmatic retargeting, mobile measurement and multichannel orchestration for most of the mid-market, and they are extending those same products into ChatGPT rather than waiting for a purpose-built ChatGPT-native competitor to take the account.
That matters because it changes what “ChatGPT advertising” is going to look like in practice for most brands. It will not, at least initially, arrive as a separate budget line requiring a separate agency relationship. It will show up as one more placement inside the media plan a marketer already runs through AdRoll, Criteo or a comparable platform, next to CTV and paid social. OpenAI itself has been explicit that initial targeting is limited to contextual signals, meaning conversation topic, general location and device type, with no personalization drawn from chat history or stored memories, and that accounts under 18 and paid-tier users are excluded by design. The ad tech layer around it is what turns that raw inventory into something a marketer can actually plan, measure and optimize against.
What it means for the marketing leader
The practical decision facing a CMO or growth lead right now is not whether to build a ChatGPT strategy. It is whether to treat this as a new line item or as an extension of tools already under contract. The AdRoll pilot argues for the latter, at least for SMB, mid-market and B2B advertisers who lack the bandwidth to stand up a dedicated conversational-AI buying function. That is consistent with a pattern this publication has tracked through the year: the shift toward conversational ads getting genuine auction infrastructure and early experiments in ad formats written specifically for AI agents rather than repurposed banner creative.
The caution is measurement. OpenAI’s stated privacy posture, no ad targeting from conversation history, no data sold to advertisers, is good for user trust and bad for the kind of granular attribution marketing teams are used to demanding from paid social. AppsFlyer and Adjust are in the pilot precisely because that gap needs to be closed with third-party measurement, not first-party dashboards controlled by the platform. Any team evaluating a ChatGPT pilot through an existing vendor should ask that vendor directly what attribution model it is using and how it reconciles ChatGPT-sourced conversions against the rest of the funnel, rather than assuming the answer is the same as it is for search or social.
What to watch next
Two things will determine how fast this moves from pilot to line item. The first is geography: the UK is live, and OpenAI has named France, Germany and Ireland as future markets pending the standard opt-in consent mechanics required under EU privacy rules, so the addressable inventory is set to expand through the rest of the year. The second is whether the early vendors report performance that holds up once budgets scale past pilot size, since “encouraging early signal” from a VP of product is not the same claim as a benchmarked cost per acquisition. For now, the more durable signal is structural: the ad tech companies that already own the relationship with the advertiser are the ones deciding how ChatGPT gets monetized, not OpenAI alone.
Source: AdRoll
