Subscribe
Share
Share
Embed
10 minutes of expert insights every weekday. Your morning ritual for staying ahead in retail media.
The Shopping Signals That Are Changing With AI
===
[00:00:00] Kiri Masters: Last week I facilitated a closed-door workshop for the Path to Purchase Institute's Retail Media Guild at their flagship Retail Media Summit in Chicago. Now, in this room, it was mostly [00:00:15] brands, but also a few agencies and tech vendors and a small smattering of retailers.
[00:00:23] And I shared three provocations for the room to discuss among themselves. And if you've [00:00:30] been listening to this podcast for a while, you'll be familiar with these. So one was dark search and how A lot of traffic that begins in an AI chat assistant is arriving at a [00:00:45] publisher or a retailer without any referral tags and is therefore likely contributing to a systemic undercounting of the volume of AI-referred traffic We talked about how [00:01:00] AI-enabled shopping
[00:01:01] may impact various formats of retail media, what that looks like across on-site, off-site, and the store
[00:01:09] And what are the five requirements for something new [00:01:15] that pops up to actually become a lasting shift? So I presented these provocations and thought starters to the room, and then at the tables, people were able to have a conversation and talk about what they're seeing
[00:01:29] [00:01:30] And what struck me was that while the conversations in the room were animated and people were excited to share and learn, the mood was overall pretty calm and curious. But calm isn't [00:01:45] necessarily the same as safe. And so in today's episode, I wanna share high-level overview of what some of these brands say they're seeing, what they are not yet seeing, at, and what is keeping [00:02:00] them up at night.
[00:02:01] Let's jump in
[00:02:02]
[00:02:03] Kiri Masters: So first of all, talking about what is holding steady for now, and so this is some areas that might be surprising in terms of how little things have changed on the surfaces that [00:02:15] people tend to be worried about the most. Number one was ad conversions. The consensus across the tables is that there hasn't been a dramatic change in the performance of ad units in terms of ad conversions yet.[00:02:30]
[00:02:30] Secondly, sponsored product ads are holding steady. There is not much erosion here in terms of performance, and nobody is reallocating in a panic. The instinct in the room is to test and monitor. No one is [00:02:45] dramatically changing any budgets just yet
[00:02:48] So next, what is actually shifting?
[00:02:51] So while the outcome of actual sales is looking steady, the path towards those sales is definitely changing. And this was the single most [00:03:00] consistent observation in the room, that the sale is still happening, but where the journey begins has changed. So there's a few different observations here.
[00:03:10] Number one, Google referrals are down. Shoppers are [00:03:15] increasingly starting in ChatGPT and Claude, and Google, at least to date, has been a little late to that game. Although I think everyone's recognizing how much they're stepping things up. Next is browse [00:03:30] traffic on Amazon has come down quite a bit by at least one account.
[00:03:35] Trust and community surfaces are rising. These are places like Reddit, Pinterest, Wikipedia. These are becoming starting points [00:03:45] instead of on-site search. Social commerce is up, and this is an interesting one, TikTok Shop being singled out as a growth area that might bypass the retail media ecosystem [00:04:00] entirely
[00:04:00] And one category that is really appreciating everything that LLMs are bringing is actually food ingredient brands. So recipe component brands reported a big jump [00:04:15] in being referenced and converted because they ride inside the meal plans and recipes that LLMs are surfacing And one nuance here in terms of D2C channels and direct channels
[00:04:29] There was [00:04:30] a little bit of a split here. It is down for some, it is up for some, and that's an interesting signal in itself that things are not necessarily evenly distributed between different product [00:04:45] categories, between different types of shopping missions
[00:04:49] Just like we use different formats and pathways today for different shopping missions, that's also the case, of course, with AI-enabled shopping journeys too [00:05:00] The agentic AI revolution is here, and the competitive edge belongs to those who move faster than disruption itself. [00:05:15] Join Miracle Ads and me in New York City on June 10, 2026, for strategies, insights, and connections built for the era of agentic commerce. [00:05:30] Discovery is shifting, media economics are following, and the brands and retailers who figure out what comes next won't be the ones who wait.
[00:05:39] They'll be the ones already moving. Join me at the Miracle Summit in New [00:05:45] York City on June 10. Link to register in the show notes
[00:05:50] And now let's move on to what we're not yet seeing. A journey that starts in AI chat is mostly invisible [00:06:00] until checkout
[00:06:00] AI can be shaping the decision
[00:06:03] But actually leaving no trace in standard analytics. Now, I've written about this before as dark search, which is the LLM equivalent of dark social. When a shopper [00:06:15] researches in ChatGPT, never clicks through to a landing page, and then buys in store or types your domain directly days later, this is what we're calling dark search.
[00:06:27] And
[00:06:28] this exists in [00:06:30] other entry points, of course. You might see an ad on TV and pull out your phone and Google that brand or that product. That exists. You might see something pop up on TikTok from an [00:06:45] influencer and just go to Amazon and search for that product instead of clicking through These types of dark search patterns exist elsewhere in consumer behavior, but with AI specifically, it's happening where a [00:07:00] conversation that happens in AI, a person might be clicking out directly to a product detail page, but that referral is missing any kind of UTM tags that you'd typically get from a search query or from Google or [00:07:15] from another source.
[00:07:17] And so this AI conversation gets zero credit. It's treated as direct traffic, and so we are just as much in the dark and having a moment of mystery in knowing where, where this traffic and [00:07:30] conversions is coming from than ever before
[00:07:33] And because measurement is more of a mystery than ever, the room started naming replacements for the metrics that no longer work. One is a new metric that one [00:07:45] participant called share of agent recommendations. And this is how often your brand surfaces and gets recommended across AI assistants.
[00:07:55] It's part of a broader shift overall towards shopper impact measures [00:08:00] over things like raw ROAS
[00:08:02] Agent readiness is now something that some brands are tracking. This is how referenceable you are in the sources that AIs cite, places like Reddit, Wikipedia, [00:08:15] and one that was new to me, brand hubs for ChatGPT. ~And finally, shopper impact measures are gaining ground. Hmm, no, just delete that one~
[00:08:21] ~I also asked the room to rank retail media surfaces by how defensible they think they're going to be against~ and I asked the room to rank various retail media surfaces by how defensible they are against these AI-enabled [00:08:30] shopping journeys. And for many tables, they put in-store retail media at the top, which is predictable enough given that 80, 90% of sales still happen there
[00:08:41] But one table had a really interesting point to make, and that was [00:08:45] actually putting influencer and word of mouth above everything else. And their logic was that it's about who you trust and why you trust them. And people do trust influencers in a way that AI can't [00:09:00] easily disintermediate. And so there were a lot of nods around the room that suggested that they were onto something here So just wrapping up here, the disruption that everyone is bracing for isn't showing up yet, at [00:09:15] least as a wholesale collapse of anything that we do right now.
[00:09:20] But it is changing. It's changing quietly. It is harder to measure, and it is easy to miss if you're just watching [00:09:30] end conversions. So maybe this is a storm that is already arriving just too softly to register on the dashboards that we've got. Maybe we're still in the calm before the storm
[00:09:41] But to me, it is the brands that are watching [00:09:45] where the journey starts, not just where it ends. Those are the brands that are watching the right thing. Thanks for listening, and I'll catch you tomorrow.
[00:09:55]