Latest / Retail Media Breakfast Club / AI Shopping Hits 75% Adoption ['Snips' from Retailgentic]
Transcript
- SNIPS 9/2/25 === [00:00:00] Kiri Masters: Let's be honest, the retail media space moves so fast, and who has time to listen to every podcast? Dropping industry insights. That's where today's format snips comes in. I'm already deep in the audio trenches, so you don't have to [00:00:15] be. Each edition of snips delivers some of the most compelling moments from industry podcasts, the aha moments, the contrarian takes, and the data drops that make me hit pause to write it down. [00:00:28] But I'm not gonna just give you the raw [00:00:30] clips. You're gonna get my take on why these insights matter, how they connect to broader trends and what it means for you. [00:00:35] [00:00:37] Kiri Masters: In today's episode, I'm recapping some of the conversation on the Retail Gentech Podcast with Scott Wino. And on this episode, [00:00:45] Scott Wino interviews Melissa Minkow, who is the global director of Retail Strategies and insights at CI and NT. Melissa shares her perspective on consumer behavior, why ingrained shopping habits stick, and how agentic commerce will shape the [00:01:00] future of retail. [00:01:01] In this part of the conversation, Melissa, is sharing exclusive findings from CITs latest consumer survey that should fundamentally shift how we think about AI adoption in commerce. [00:01:12] Scot: Have you done any consumer research on the gen [00:01:15] AI side? [00:01:15] Melissa: We've done a ton, actually, like literally just finished fielding this survey that has a ton of information on that. , So on generative AI in particular, consumers are definitely using it way more frequently. Some of the statistics that really stand [00:01:30] out to me, , practically three quarters of consumers said they're using AI tools in their path to purchase at least occasionally. [00:01:36] Only one quarter of respondents said they have not used an AI tool during their path to purchase [00:01:40] Scot: I've, [00:01:40] seen 60% I think is the highest. Um, so three quarters would be a new high watermark, there [00:01:45] hasn't been a lot out since GPT five dropped in early August, so I don't know when you ran your survey, but [00:01:49] Melissa: yeah, no, this is fresh, fresh data. This was done fielding, I wanna say, end of August last week. So, yeah, I'm proud that this is incredibly recent and I do think it's. [00:01:58] Very accurate because our sample [00:02:00] size is a thousand US consumers and they are reflective of the population. It's not skewed younger [00:02:04] [00:02:06] Kiri Masters: 75% of consumers, wow. That is not just incremental growth and the timing matters here. This data, as Melissa said, was fielded [00:02:15] in late August, making it some of the most current consumer research available. And importantly, that sample reflects the broader population, [00:02:22] not just the early tech adopters like you and me. [00:02:25] Now let's move on to another part of the conversation where Melissa and Scott weighed into [00:02:30] the perennial complaint about Amazon's ad saturation. [00:02:34] Scot: Any other kind of things in the report that shocked you? [00:02:36] Melissa: I was very surprised to see that. The primary way consumers feared AI would lower the quality of the shopping [00:02:45] experience was by having some sort of agenda in terms of what it shows to them. [00:02:48] So this is really important because I know you were going to ask me my thoughts on almost like sponsored GEO. And retailers being able to pay either to play or to be higher up in the [00:03:00] results because there's not going to be a second page anymore, et cetera, et cetera. I really caution against these third parties doing that because it will completely erode their credibility along the path to purchase. [00:03:10] And I think Amazon is a great example of the fact that consumers have trained themselves to [00:03:15] not just click the first thing that they see. We're really skilled. We're really well-informed shoppers these days, and. If we think we're being sold something because a brand has paid and it doesn't live up to that, like maybe we'll try it first, but if it doesn't live up to that, they [00:03:30] wise up really quickly and they'll change their shopping behaviors. [00:03:32] Scot: Yeah, we had this professor on and he had data that showed people trust gen AI more than a friend. Um, and then I saw another one. This was an eMarketer where it's, uh, website is at like 9% and Gen AI is at 18% just kind of tied with store at [00:03:45] this point. [00:03:45] So it's almost tied with going into the store, seeing the product and talking to a sales associate. I think Amazon's totally jumped the shark on ads and. I get retail media networks, and I think they work okay, sprinkled in what they actually add value, but they've all tipped over into this like, eroding trust and the experience.[00:04:00] [00:04:00] Melissa: Exactly. And if you overpower the experience, then the consumer learns to tune you out. So it doesn't behoove you to have. So much content being thrown at the customer all the time, or so many ads, or so many brands all the time. I mean, to your point, the credibility right now and the trust is there. [00:04:14] So it [00:04:15] would be foolish to lose that opportunity by allowing so many brands to pay and be in that space that way. [00:04:20] [00:04:21] Kiri Masters: Okay, so here is where I respectfully disagree with both Melissa and Scott and sort of part ways with the conventional wisdom. [00:04:28] Yes. Shoppers [00:04:30] complain about Amazon ads and there is a lot of Amazon's real estate taken up with ads, but the complaints here don't equal the business impact and the actual behavior that's happening. Amazon's advertising revenues still [00:04:45] growing very quickly, and a lot of that comes from sponsored products and bottom of funnel, onsite advertising and the brands that are. [00:04:54] Spending money on onsite ads are continuing to invest because like it or not, [00:05:00] as a consumer, those ads deliver measurable returns. Sure you're crushing it. Sure. You're crushing Sure. You're crushing it on Amazon and Walmart. But what about Kroger, [00:05:15] Instacart, and the fast growing regional grocers? [00:05:18] A Costa Group builds holistic media plans on real shopper behavior across all the US retailers that your brand cares about. from digital shelf to retail media to [00:05:30] data analytics, Acosta Group offers end-to-end connected commerce all under one roof. Learn more at ACOA Group and tell them Retail Media Breakfast Club sent you. [00:05:42] That's a C OS T [00:05:45] group. [00:05:48] A couple of months ago, I recapped an academic study analyzing over 120,000 brands on Amazon and how the different ad formats work and which ad formats. [00:06:00] Perform the best I am so shocked how little this study is actually, discussed. Definitely go check this out if you are involved with the Amazon ads ecosystem. [00:06:10] But the long and short for the purpose of this conversation is all of the different ad [00:06:15] formats on Amazon consistently drive awareness, consideration, and revenue across all different sizes of brands. Ads on Amazon work. So I think the real issue here isn't necessarily ad volume. It's ad [00:06:30] relevance. [00:06:30] Whatever activity you're engaged with, if you see an ad that's irrelevant, you notice it. If you see an ad that's relevant and it's from a category that you're in market for, or a brand that you already like, and they're introducing a [00:06:45] new colorway or scent or flavor or whatever. That feels like a helpful experience. [00:06:51] So this is where I think, you know, there's a lot of dumping on Amazon's ad saturation, but you can see from the numbers that they continue to [00:07:00] work. All right, my small rant, they're over. The conversation then shifts to some more interesting research about AI shopping behavior as it relates to sponsored products. [00:07:11] Scot: There was another research report, this one was from university. I can't remember if this was [00:07:15] Cornell or Stanford or one of those. And, um, they built pretty generic agents using just open AI and standard things and gave 'em basic prompts like. [00:07:21] Go find this product. And without prompting the agents would avoid sponsored listings and advertising and things like that. So it's interesting. I don't know if they're parroting consumer [00:07:30] behavior or they kind of intuitively know, um, also that frequently that's a red herring of it's not the best deal, not the best product on the ads. [00:07:36] [00:07:37] Kiri Masters: Scott's referring here to a study by researchers at Columbia and Yale and an enterprise AI software company [00:07:45] that I covered extensively a few weeks ago. And what it found was AI agents penalize sponsored products by eight to 14% in selection probability. So the AI agents are avoiding products that are sponsored, and it [00:08:00] doesn't seem like the agents were programmed to avoid ads, they just organically developed this bias. But here's the thing about that AI model behavior. It's trained on human generated content where we explicitly express distrust of advertising even [00:08:15] when our actual behavior suggests otherwise. These current AI models have learned our stated biases. Not our revealed preferences. [00:08:23] And that's why I think this won't persist as the AI platforms compete for users, they'll need to deliver [00:08:30] genuine value. Sometimes that genuine value is the sponsored product offering that is the best deal or the best solution. And so an agent that categorically dismisses. [00:08:39] Sponsored items is gonna miss legitimate opportunities, and that will be frustrating to users who [00:08:45] discover that there was a better product for them and that the agent just disregarded it because it was sponsored. [00:08:50] More importantly though, wherever consumers spend time advertising follows, the major LLMs will eventually roll out advertising models because the economics [00:09:00] demand it. People say they don't like ads. You know what they don't like even more is paying a subscription fee for something, right? So the AI platforms, especially the biggest ones, are gonna find ways to monetize attention while [00:09:15] attempting to maintain user trust as well. [00:09:17] ~All right, let's move. Let's go back to the conversation between Melissa and Scott, where Melissa's. Where Melissa's gonna talk about her framework for thinking about the different AI agents when Melissa provides a framework for thinking about three different types of AI agents.~ [00:09:17] ~All right, we're gonna go back to the conversation between Melissa and Scott. Melissa's gonna talk about the different types of AI shopping agents.~ [00:09:17] Let's move on to Melissa's framework for understanding three different types of AI agents. [00:09:23] Melissa: Exactly. And that goes back to kind of the three types of agents. If it's a third party that is now eroding its own credibility, that's not [00:09:30] great. If it's the consumer themselves building their own agents, they will teach them and build them so that they avoid those pitfalls. If it's the retailer's agent, I think there's a really unique opportunity there to treat it like the rest of the website and obviously promote what makes sense to be promoting [00:09:45] through deals, discounts, et cetera. [00:09:46] But. Be aware that the consumer has a different mindset with your agent than they do with their own agent, because they're inherently already on your website. Inherently kind of biased. Either towards you or against you, because [00:10:00] they're just so hyper aware of where they are. [00:10:01] Scot: Yeah. And by having your own agent, you mean like, uh, rufuss and Sparky and those types of guys, kind of the branded retailer inside the walled garden type of a thing? [00:10:08] Melissa: Exactly. And if it's a direct to consumer retailer, them creating their own as well. [00:10:12] [00:10:13] Kiri Masters: I think this is a really interesting point [00:10:15] around contextual awareness. When shoppers interact with Rufuss, they understand they're an Amazon's ecosystem. Rufuss is only gonna recommend products in Amazon's ecosystem. [00:10:26] The challenge for retailers is gonna be around balancing promotional [00:10:30] opportunities with maintaining the trust that makes these shopping assistance valuable in the first place. So wrapping up my thoughts here. We're seeing these really high AI adoption rates and they seem to be climbing higher and higher as we draw the year to a [00:10:45] close. [00:10:45] Consumers are becoming more familiar with these tools. They're using them more frequently, so retailers and brands need to be thinking through what does this mean? [00:10:53] Not just now as adoption ramps up, but in the future when a lot of this web traffic or app traffic that. [00:11:00] You are seeing is gonna come in a lot more informed about what they want and may have already made their mind up about what they're going to to buy that day. There might be less discovery and browsing and consideration happening on site because at [00:11:15] least some of that is happening. [00:11:16] Upfront in chat GPT or perplexity or wherever a consumer is spending their personal time. [00:11:22] Big shout out to Melissa Minow from CINT sharing a sneak preview of this data that they're going to be dropping in a [00:11:30] couple of weeks. She did say in the episode the best place to find that research when it drops is on the CINT website. That's CI. A NDT and also check out the Retail Gentech podcast. [00:11:43] Scott's a great host with lots [00:11:45] of viewpoints of his own, but he's bringing in a ton of great guests recently, so check that one out too. Thanks for tuning in, and I'll catch you tomorrow. [00:11:52]