AEO will dictate how well your ChatGPT ads will perform

Abuyer asks for three suppliers. Two get named. Before spending to reach that buyer, I want to know who vouched for those two and why my brand missed the cut. That’s where KXP and AIRE belong. KXP traces the question, the recommendation and the source behind it. AIRE uses that evidence to fix what buyers can find on the brand site, at retailers and through people they already trust. Then ask again and see whether the shortlist moves.

The money is already moving. On August 31 OpenAI put its advertising business above $1 billion in annualized revenue, a milestone it said it reached in less than 200 days. That is the pace at which advertisers were buying inventory. It tells a marketing director very little about what the next dollar will buy for their own brand. OpenAI’s announcement

The half inch I’d put on the media brief

Take a question from our own fieldwork. A traveler wanted a 45-litre bag for camera gear that could travel in the cabin. ChatGPT recommended Peak Design’s Travel Backpack 45L. If I sold that bag, I’d be pleased with the mention. Then I’d ask the question the person packing a camera is likely to ask next. Can I bring it on Delta? Kitsune’s field report

Peak Design lists a standard depth of 9.5 inches. Delta gives a maximum of 22 by 14 by 9 inches, including wheels and handles. Half an inch on the depth is enough to change the copy I’d approve. A soft bag may pack down. How much room will the camera leave to do that? Who wants their first answer to come from a gate agent with boarding underway? Peak Design’s bag · Delta’s cabin rules

That half inch is what the organic answer earns the paid campaign. The customer came looking for a way to travel with equipment. A search visibility report can tell the brand it won the mention. A good AEO team can tell the brand which claim to retire before it puts money behind it, which photograph needs a tape measure, and which buyer should be shown a smaller bag. I would take that meeting before the media plan gets signed.

What the ad buyer inherits from the AEO team

Paid search people learned this relationship years ago. A landing page that does a better job of answering the search gives the person buying the keyword a better place to send the click. ChatGPT makes the relationship more interesting because the buyer can add a condition halfway through the conversation. “I’m taking camera gear” gets you into the running. “It has to fit on Delta” can take you straight back out.

OpenAI describes two systems. ChatGPT writes its answer independently of the advertiser. The ad system considers the conversation, the ad title and copy, the destination page and the bid. Its tools can even use website metadata to draft an ad. Both teams at the brand end up drawing on the same product page. If that page says only “travel-ready”, what useful distinction is the media buyer supposed to make between the weekend traveler and the photographer flying Delta? How ChatGPT selects ads · OpenAI’s ad creation guide

This is where I’d use KXP, our check on what AI recommends. Save the actual question, the product picked, the alternatives and the source that settled it. Give the product team that record through AIRE, our AI Recommendations Engineering program. For this bag I want dimensions that specify how it was packed, airline limits clearly attributed to the airline and a link to a smaller option if the fit is tight. Then ask again. If the recommendation changes when you specify the airline, you’ve learned something about two buyers that a brand-level visibility score would have averaged away.

After the click, the buyer can ask again

OpenAI’s Sponsored Business Agents let eligible advertisers open a product conversation from an ad. The setup guide lets a business draft its agent from its website or FAQ, then test likely customer questions. Imagine paying to reach the photographer twice in one session, first through the ad and then through an agent whose entire knowledge of the bag amounts to “travel-ready”. The expensive bit was getting the right customer into the conversation. The cheap bit was publishing the measurements. OpenAI’s Sponsored Business Agent guide

Give the improved page and the current page comparable traffic. Track what it cost to win an order, how many buyers asked support about cabin fit, and how many sent the bag back. If the more specific copy earns fewer clicks and more orders survive the returns window, I know which campaign I’d scale. A marketer can buy another impression this afternoon. They should already know what their brand will say when the customer asks whether the thing fits.