Latest / Retail Media Breakfast Club / The Retail Media Illusion: Why Agentic Commerce Will Disrupt E-Commerce Sooner Than You Think
Transcript
- The Hallucination Is That Everything Is Just Fine As It Is === [00:00:00] Kiri Masters: Last week I shared a stage with Andrew Lipsman and Rob Gonzalez at the Digital Shelf Summit in Atlanta. [00:00:07] It was a debate with the motion that agentic commerce will be an existential crisis for e-commerce traffic. [00:00:15] I argued the affirmative in an admittedly extreme position, and Andrew and Rob argued the negative. And a lot of the ground that Andrew covered in his talking points were [00:00:30] covered in his guest post in Debbie Aho Williamson's AI Ad Economy newsletter a couple of weeks ago. [00:00:37] And I saw this come out, this guest post come out in Debbie's newsletter, and I asked her if I could write a [00:00:45] response, a rebuttal. And so that piece is gonna run in her newsletter next week, and if you're not subscribed to the AI Ad Economy, you should definitely do so. It's great companion reading to this podcast. [00:00:57] But today I'm gonna share a version [00:01:00] of that rebuttal with you, dear listener. And unfortunately, I'm gonna have to cut a few things short in the audio version. There's a lot going on in this essay. If you wanna read the [00:01:15] full argument, certainly check that out. We'll link up to it in the show notes. But for now, let's jump in and see how much ground I can cover in 10 minutes. [00:01:24] [00:01:25] Kiri Masters: So first of all, for those who haven't read Andrew's piece [00:01:30] about agentic commerce being a collective hallucination, I'll give you the short version. [00:01:37] he says this is a story that the industry has told itself into believing. It's fueled by echo chambers, [00:01:45] management consulting firm prognostications, and headline-grabbing statistics that don't survive scrutiny. Here are the three Core pillars of his claims. [00:01:57] First, he talks about this, um, [00:02:00] traffic figure from Adobe that shows that AI referral traffic has grown six hundred and ninety-three percent, and he explains that that traffic number is coming off a tiny base, [00:02:15] and so such a huge traffic growth number doesn't really stand up when you look at what the denominator is. [00:02:22] Second is the viral chart that claims ChatGPT is the number two e-commerce site be-besides Amazon, [00:02:30] and he shows how the math was cherry-picked, uh, around which sites to include. For example, n- they didn't include Google in that, and that ChatGPT's actual product-oriented traffic is really, really small when we c-compare [00:02:45] that to every major retailer. [00:02:46] And then finally, the agentic commerce is an existential threat to retail media networks. This is me that he's talking about amongst a few others, and he pulls eMarketer forecasts [00:03:00] showing that commerce media ad spend will grow forty-eight billion dollars through twenty twenty-nine versus twenty-eight billion dollars in agentic commerce sales. [00:03:12] And he uses this to argue that [00:03:15] retailers should keep building their media networks and stop worrying about this distraction. So I wanna give Andrew his due on points one and two. Those numbers from Adobe were over-indexed [00:03:30] on. We didn't get to see what that denominator was from their calculations, and the ChatGPT chart was bad math. [00:03:40] But I argue that on point three, that agentic commerce is an [00:03:45] existential threat for RMNs, that he is measuring the wrong thing. So Andrew's argument is a forecast about transaction share by twenty twenty-nine. But the argument that I've been making for the last few months is about where customers [00:04:00] do their narrowing. [00:04:02] A forecast can tell you how much money might move through agentic commerce transactions in three years' time, but it's not telling you what buying decision is being made today, and that's the part that retail media's business model [00:04:15] actually rides on. Retail media doesn't make money directly from transactions. [00:04:21] It makes money from impressions on the surfaces that consumers visit on their way to a transaction. And as it stands today, for better or [00:04:30] worse Most of the ad revenue from retail media is from on-site placements, search result pages, category browse pages, homepage banners. Those surfaces depend on consumers showing up to do the work of discovery, [00:04:45] comparison, and decision-making on the retailer's site. [00:04:49] And if that work moves upstream into an AI assistant, the transactions might still happen on a retailer.com, but the retail media [00:05:00] impressions go missing. And that's the argument I've been writing about for nearly a year now. Andrew's argument doesn't address it because he takes umbrage at the conflation of the terminology of AI-assisted shopping and [00:05:15] full-fledged agentic commerce. [00:05:16] But I'd call that a distraction from how emerging consumer behavior could vastly reshape the economics of retail media. [00:05:30] Miracle Ads is the Ad Tech solution trusted by Rakuten and over 50 global enterprise retailers. That's because Miracle Ads was built with both three P Marketplace sellers and one P suppliers in mind. [00:05:45] Both advertiser audiences demand a seamless advertising journey from onboarding to reporting. [00:05:51] You can offer everything from sponsored products to video ads all in one solution. Learn more@miracle.com. That's [00:06:00] M-I-R-A-K l.com. [00:06:03] So now I'm gonna turn to my sources on this topic, including one that Andrew has relied on himself. First is Criteo. Criteo recently published their [00:06:15] twenty twenty-six Commerce and AI Trend Report, and their top-line view is that AI is additive, that buying stays centralized at trusted retailers, that the hype overstates what's happening. [00:06:28] It almost reads like an [00:06:30] endorsement of Andrew's piece. But their own network data tells a sharper story here. Across roughly five hundred Criteo merchants in the US in January and February of this year, more than seventy [00:06:45] percent of LLM-referred users now land directly on product pages. In mid twenty twenty-five, that figure was fifty percent. [00:06:56] So in roughly six months, the share of AI-referred shoppers [00:07:00] who skip everything else on the retailer's site and go straight to the PDP went from half to more than seven in ten. Now, Criteo frames this as good news, and because from a pure e-commerce P&L [00:07:15] point of view, they're seeing higher conversion, less friction, more shoppers arriving with intent, so that's great. [00:07:20] But for retail media specifically, that pattern is the problem. Second source here, Adobe's quarterly [00:07:30] AI traffic report from April of this year makes the same point from a different angle. AI-referred visitors spend forty-eight percent more time on retail sites, view thirteen percent more [00:07:45] pages, and convert forty-two percent higher than non-AI traffic. [00:07:50] Now, if you think this through logically, an AI-referred shopper arrives with a short list. They might open [00:08:00] three PDPs to compare reviews, specs, price, make sure that they're making the right decision. [00:08:06] They pick one, they check out. Now, that's 12 pages that they just viewed and 15 minutes on the [00:08:15] site. What they don't do necessarily is browse the category page, do keyword s- searches, review the results there, look at the homepage, or click into the deals section. And those are the surfaces where [00:08:30] on-site retail media's premium inventory lives. [00:08:34] And so this is the version of the future that I've been describing. Engagement might be going up, but on-site ad inventory and impressions is going down. [00:08:44] Now, I have [00:08:45] two more arguments in my piece today. I'm not going to be able to do justice to both of them. I'll just give you a quick preview. One objection that I hear a lot is that upstream influence isn't new. We've been seeing upstream [00:09:00] influence happen with TV ads, radio, social media. But the difference here is [00:09:06] That these ads influence brand preference. Maybe they drive awareness, maybe they widen the funnel. But [00:09:15] AI assistants get so granular down to the product level, which SKU you should buy, what-- which retailer has the best price. All that narrowing is done in the AI assistant rather than at a [00:09:30] retailer. [00:09:30] And so that's what we're seeing in the wild with this direct-to-PDP shopping behavior. I also make the argument that people are already exploring really janky, duct-taped-together [00:09:45] versions of this shopping activity because not all shopping is fun. Some of it is fun, some of it is tedious, repetitive, annoying, and we would like to outsource it, and people are [00:10:00] starting to do that. [00:10:01] And we've seen through history that sometimes the behavior precedes the future business model, and we are seeing people develop these systems using [00:10:15] scripts and Open Claw and things like that to escape the tedium of some shopping behavior. Wrapping up here, Andrew closed his initial piece with a warning. [00:10:26] Don't be surprised when you get burned. And I would hand [00:10:30] that right back to him. The hallucination that he's worried about, the one where the industry talks itself into agentic commerce as imminent disruption, that might be real. But the bigger hallucination is the belief that nothing about the current system needs [00:10:45] to change, that the surfaces that retail media is built on will keep generating the same impressions at the same volumes and prices for a journey that looks like it did in 2023. [00:10:57] Andrew is right that the headlines might be [00:11:00] overselling imminent disruption from agentic commerce, but he is wrong that retailers should blindly continue with the same playbook they've been running for the past five years. The hallucination isn't that agentic commerce is coming. [00:11:14] It's [00:11:15] the belief that nothing about the current system is immune from change. [00:11:19]