Beyond Retail Media: Connected Commerce with Pacvue’s Tommy Burton

Full analysis

Amazon put its ad-buying software behind ChatGPT. On a Pacvue-sponsored, Amazon-sponsored podcast, Tommy Burton, VP of Global Partnerships at Pacvue and a former Amazon Ads executive, flagged a just-announced deal that lets brands buy ChatGPT ad placements through the Amazon DSP. That is the one piece of real news buried in an hour of product showcase. Everything else, the $70 billion retail media figure, the HP case study, the AI agent, is Pacvue talking its own book on a platform its book depends on.

So let's separate the signal from the sponsorship.

What's being decided: Nothing you can undo or not undo. This is a briefing, not a build decision. The live question for operators is whether the Amazon-ChatGPT tie-up changes where commerce budgets flow and who gets to measure them. Easy to wait on. Nothing here forces a move this quarter. The one thing worth acting on early is measurement infrastructure, because that window closes as retail media consolidates.

The Market Analyst. Follow who owns both the data and the surface. If Amazon DSP inventory now shows up inside ChatGPT answers, Amazon has extended its buying rails into the fastest-growing discovery surface on the internet, and it did so before Google, Meta, or The Trade Desk had a product there. In plain terms: when you ask ChatGPT what laptop to buy, Amazon wants to sell the answer. That favors Amazon and Walmart Connect, who close the purchase loop natively. It squeezes independent measurement, because you cannot audit a conversion Amazon books inside a chat window it also monetizes.

The Skeptic. Burton presented the ChatGPT-DSP deal with no date, no source, and a direct commercial interest in you believing it. Pacvue makes money when advertisers pour more into Amazon DSP. "All media is retail media" is a slogan that happens to describe exactly what Pacvue sells. And the HP numbers, 29% more reach, 100%-plus conversion lift, 500 hours saved, are self-reported, on an Amazon-sponsored episode, with no baseline, no time period, no independent audit. Every headline number here comes from someone paid to make it sound big.

The Operator. The useful admission is that Amazon Marketing Cloud, the most capable clean room in retail media, is "highly underutilized." That is a confession that brands cannot answer the one question that matters: is this spend incremental, or am I paying to capture sales I already had? Burton calls that the "ad tax." He is right. Any agency that builds a working AMC practice, or licenses a tool that queries it in plain English instead of SQL, has something clients will pay for today. That is real work with a real deadline, because it gets harder once the networks consolidate their measurement on their own terms.

The CFO. Retail media at $70 billion sounds like opportunity. Read it as cost. Much of that spend is defensive, brands buying their own branded search terms so a competitor does not. The incrementality question is the whole game, and the platforms that sell the media also grade the homework. Amazon extending into ChatGPT does not change that math. It makes it worse, because now there is a new surface where the seller reports the score and no third party can check it.

Where the council splits. The Analyst sees a genuine structural move: Amazon first into agentic shopping. The Skeptic sees a salesman flagging demand drivers that fatten his own platform. Both can be true. Amazon really did the deal, and Burton really has every reason to oversell how urgent it is for you.

What this hinges on. One fact: does commerce conversion data become the currency that other channels get measured against? If yes, the walled gardens that own both the transaction and the media win, and open-web publishers plus independent measurement firms lose the right to grade their own value. If measurement stays fragmented and no single conversion signal wins, the status quo holds and this is just another buying surface.

The council leans toward Amazon gaining, but distrusts the framing. Verify the ChatGPT-DSP deal independently before you brief a client on it. Build AMC muscle now regardless, because incrementality measurement is the gap that pays, and it pays whether or not the chatbot ads pan out.

Prediction: By 30 June 2027, advertisers will not be able to buy ad placements inside ChatGPT through the Amazon DSP as a product open to any Amazon DSP advertiser.

Confidence: Medium

Why: The claim that Amazon's demand-side buying platform now reaches ChatGPT placements has circulated through vendor briefings with a commercial interest in more Amazon DSP spend, without a dated confirmation from either Amazon or OpenAI. OpenAI is building its own advertising business and its own shopping flow, and routing that inventory through Amazon would hand a retail competitor the pricing and the conversion data on the exact moment a shopper decides what to buy. On the buy side the demand is soft: Amazon Marketing Cloud, the most capable clean room in retail media, sits underused on Amazon's own mature surfaces, so advertisers already cannot tell whether that spend is incremental, and a chat surface with no shared identifiers and no agreed conversion definition is a harder sell, not an easier one. For the opposite to happen, two companies racing each other for the same commerce transaction would have to ship a joint, generally available buying path inside nine months.

Revisit by 2027-06-30: We're right if On 30 June 2027, Amazon Ads' public product documentation does not list ChatGPT or OpenAI inventory as a placement any Amazon DSP advertiser can buy. We're wrong if On 30 June 2027, Amazon Ads' public product documentation lists ChatGPT or OpenAI inventory as a placement any Amazon DSP advertiser can buy.