Latest / Retail Media Breakfast Club / AI Is Clicking Your Ads (And Wasting Your Budget) — The Hidden Retail Media Crisis
Transcript
- No Eyes on the Ad === [00:00:00] Speaker: the headline post on the news website, business of Fashion last Friday was how AI shopping agents are reshaping fashion's ad economy. And it was a piece about how AI agents are distorting ad [00:00:15] metrics. Now I was asked to contribute to this piece and I was quoted on a bit about emotional storytelling that as AI search shortens the time that shoppers spend on brand sites, the moments when a human [00:00:30] actually does show up themselves matters more. [00:00:33] But reading the rest of the piece and insights from new research that was cited here made me realize that I've been missing something in my own coverage. ~I've spent some time now arguing that a agentic commerce is a structural threat to retail media with ad surface, with ad surfaces disappearing first party data, losing some of.~ [00:00:43] ~First party data, losing some signals. The three pillar economics of RMNs.~ [00:00:43] I've spent some time now [00:00:45] arguing that ag, agentic commerce, and even just AI enabled shopping is a structural threat to several aspects of retail media, including ad surfaces disappearing for onsite ads and [00:01:00] first party data, losing some important behavioral signals. [00:01:04] What I hadn't focused on is a more immediate and messy problem that AI agents are already out there clicking around. [00:01:14] And [00:01:15] brands are paying for it. [00:01:16] [00:01:17] Speaker: So here's how it works. An AI agent navigates to a retailer's website. And they could be looking up things like product availability, checking prices, trying in some cases, [00:01:30] to actually add something to a shopping cart. Now, the retailer's site can't tell if this visitor is human or not. [00:01:39] The agent's activity triggers some ad impressions. [00:01:44] And at least [00:01:45] under a CPM model, the advertiser gets charged, but no human ever actually saw that ad. Now this is the logical endpoint of a problem that. Jason O'Toole, who is the head of Connected Commerce and media [00:02:00] at Apparel Brand Gildan was describing just last week. In an interview, he said the quality of impressions isn't standardized, and a lot of the times if you scroll past an ad, the consumer may never see [00:02:15] it, but it will still count in certain platforms as an impression. [00:02:18] That is a problem that we have to today with impression based ads, but at least in this case, a human scrolling past an ad. Could have seen it, but a [00:02:30] bot has no eyes. Now this isn't hypothetical anymore. Business of fashion sites, cybersecurity firm, human security, who say that a agentic traffic grew by more than 1300% [00:02:45] in the first eight months of 2025. [00:02:47] It's still a small share of overall web traffic. With only a very small amount of users using agents for checkout right now, but if that [00:03:00] behavior continues to compound, [00:03:02] the ad industry's measurement infrastructure is not ready for this. [00:03:06] And double verifies. VP of Product Management described a lab test where an AI assistant tried to buy shampoo that [00:03:15] was out of stock. Rather than give up and move on, the agent kept clicking. 10 times, 20 times working through different sections of the site, trying to complete the task. Boy, they're diligent. [00:03:29] [00:03:30] Every one of those clicks is a potential chargeable ad event, but the ad spend problem is really only half of it. I was just thinking about this as I read the Business of Fashion article that these clicks that [00:03:45] an AI agent is making. They're not just triggering impressions. [00:03:49] They are feeding the retailer's behavioral data, pool, search queries, product page visits, category browse patterns. All of this, [00:04:00] these behavioral events that are largely driven by clicks get logged as member intent or visitor intent. And for a retailer with a logged in user who has an agent working on their behalf, [00:04:15] every one of those fruitless clicks now looks like genuine consumer interest that that member. [00:04:21] Now appears to be a heavy shampoo shopper, desperate to find the right product, and [00:04:30] that user gets updated in that audience segment. Brands now buy against that audience segment. The targeting is wrong from the very start, [00:04:40] So agent traffic doesn't just waste ad dollars. It actually [00:04:45] poisons the data pool that justifies both onsite and offsite ad dollars. Did you know that leading retail media [00:05:00] networks drive 85% of their ads through mid and long tail advertisers? [00:05:06] Kiri Masters: Miracle Ads provides full funnel ad formats tailored to both one P and three P advertisers leveraging unique [00:05:15] AI capabilities that provide unprecedented levels of relevance and engagement. Retailers who want to capture ad spend from the long tail of three P Marketplace sellers use miracle ads in their tech stack. [00:05:29] Learn [00:05:30] more@miracle.com. That's M-I-R-A-K l.com. [00:05:36] [00:05:36] Speaker: The response coming from companies like human security and Double Verify is something called cost per human. Essentially only [00:05:45] billing brands when a verified human viewed or clicked an ad. So that sounds kind of obvious, but the catches that distinguishing human from agent traffic is genuinely hard. [00:05:56] And so what does all of this mean for retail media [00:06:00] networks specifically? Here's a few things worth watching. Number one, the measurement problem that we have today just got harder. RMNs are already under pressure on attribution and incrementality. Now, add a more [00:06:15] basic question. Are your reported impressions and clicks coming from humans at all? [00:06:20] And harder still. Is the behavioral data underlying your audience segments accurate, or has it been quietly corrupted by agent [00:06:30] activity [00:06:30] Number two, cost per human could be good for the industry, but painful in the short term. If this takes hold as a billing standard, it could force a more honest accounting of real consumer attention, but it would almost certainly [00:06:45] reduce current reported metrics, which is uncomfortable for any RMN trying to justify its ad rates to brand partners. [00:06:55] And number three, the data problem is harder to fix. Then the billing [00:07:00] problem, you can change how you charge for impressions, but cleaning up a corrupted behavioral data pool is a different challenge, especially if you don't know how long agents have actually been muddying the waters. [00:07:15] I've been covering the existential long arc version of this story. [00:07:19] What happens to retail media when consumers stop showing up on retailer sites at all, or showing up much later in their shopping journey? But this here is the near term [00:07:30] version. What happens when the visitors showing up aren't consumers at all? Both problems are real, but they are on different timelines. [00:07:39] Now just before you go, I wanna make sure that you know about a webinar that I'm doing this week [00:07:45] with In-Store Marketplace and Catalyst Consulting. [00:07:48] It is all about in-store retail media and are we measuring the right things? [00:07:55] And the crux of it is this. Has the retail media industry [00:08:00] skipped basic maths and gone straight to advanced calculus. When it comes to in-store retail media measurement, we've rushed into some complex attribution models and digital style measurement for in-store [00:08:15] media. Before anyone agreed on the fundamentals, and that means that brands, retailers, and agencies are all working from different scorecards, and that is making it harder to grow in store investment. [00:08:28] Not [00:08:30] easier. This is. this live stream on LinkedIn is based on new research from in-store, marketplace and Catalyst Media, and it's gonna be a great discussion. So please join me on Wednesday at 11:00 AM Eastern time. We'll link up to it in the [00:08:45] show notes of this episode. [00:08:47] Thanks for listening, and I'll catch you tomorrow. [00:08:49]