Latest / Retail Media Breakfast Club / The Shortcut Demon: The AI Data Problem Nobody In Agentic Commerce Wants to Fix (Demons Series Part 3 of 3)
Transcript
- Part 3 - The Shortcut Demon === [00:00:00] Kiri: Welcome back to the Retail Media Breakfast Club podcast. I'm Kiri Masters, and today I'm joined by Anne Hallock, VP Americas for Miracle Ads. Welcome back, Anne. [00:00:10] Anne: Kiri, thanks for having me. [00:00:12] Kiri: So we've been doing a series the last couple of weeks [00:00:15] talking about how the biggest threats to retail media this year aren't external, they're internal. [00:00:21] There's three structural problems the industry has been avoiding. They are the demons inside of retail media. week one we talked [00:00:30] about the growth demon. Week two, we talked about the silo demon. And today we're talking about the shortcut demon, which is everyone wants the AI upside, the agentic commerce, traffic, the LLM driven [00:00:45] discovery, autonomous shopping agents. [00:00:47] But the prerequisite for any of that is foundational data work, catalogs, taxonomies, attributes, pricing accuracy. Well, that doesn't sound very fun.[00:01:00] [00:01:02] And the AI initiative fails. [00:01:04] [00:01:05] Kiri: So Anne, at Possible just a few weeks ago you were talking about, "Hey, don't assume that just because we have AI that everything will be [00:01:15] easier by April of next year." What does that mean? [00:01:20] Anne: That's right, Kerry. We were at Possible in Miami. Everyone who lives in the Northeast was so happy to be down in the sun. [00:01:27] Mm-hmm. I live in Texas. I am [00:01:30] immune to the charms of Miami. It feels very similar to Texas. Um, but you know, I think one of the things that we've seen, and my heart sort of breaks for people, which is there is a sincere effort for [00:01:45] retailers to understand the implications of AI, and there is a rush for everyone to claim that they are doing something with it or using it. [00:01:58] Kiri: Mm-hmm. [00:01:59] Anne: And the [00:02:00] next question that every single person should ask when a leader publicly says, "We're partnered with some LLM," is to ask the question- Mm-hmm ... "Well, how? How are you partnered with them? What are you actually doing?" [00:02:14] Kiri: Details, [00:02:15] details. [00:02:15] Anne: Details, details. Because the, the biggest challenge, and I, I feel blessed to be sitting at the center of gravity for e-commerce. [00:02:25] You know, with Miracle- Mm ... we power more than 450 retailer marketplaces [00:02:30] worldwide. And we saw this coming a couple years ago, so we actually have a chief AI officer, Anne-Claire Bouchet. We have a head of agentic commerce, a woman named Emil- Amelia Van Camp, who you've actually had on the podcast before. On this [00:02:42] Kiri: show. [00:02:43] Anne: Really fantastic- Yeah ... [00:02:45] conversations about the idea that the product detail page is the new homepage, uh, which was a super fun episode that I've shared with a bunch of different people following conversations where it comes up. [00:02:55] Kiri: Mm. [00:02:56] Anne: Uh, but really the education cycle For how to get it [00:03:00] right is a long one, which is that we end up first talking about data hygiene and the infrastructure that's required to surface meaningfully inside of LLMs. [00:03:10] And so if you think of an LLM as your clone, you are now [00:03:15] delegating someone with the power of discernment to go shop for you. You really have to think about what attributes are actually going to surface in the shopping cycle that are going to matter. So if you say, um, "I'm going to Cannes soon, I need a, you [00:03:30] know, maxi length linen dress that has to have good reviews, and it needs to get here by next Friday." [00:03:38] And you delegate that out, the LLM has to go look for products that have those attributes assigned to it. [00:03:45] Well, those attributes- Mm ... are unassigned if someone on the data side didn't say, "Hey, I'm gonna push expected delivery date. I'm going to push- Mm ... length of dress. I'm going to push, you know, color of dress, fabric of dress, et cetera." [00:03:58] And so the way that we have [00:04:00] been, uh, that we have become accustomed to shopping is really just being replicated, but it's being replicated in a system where it has to be able to draw from information that needs to be provided by [00:04:15] someone. And so we get into this moment- Right ... where Mm-hmm ... you start having a conversation about agentic AI, and I, you know, I was talking about this idea of, like, we were all promised flying cars. [00:04:27] Yeah. And you start talking about data [00:04:30] hygiene, and you just see the light die from people's eyes. Like, it just goes out. Because it's not a fun thing to have to talk about eating your vegetables if you wanna be healthy, and it's not a fun thing- Yeah ... to talk about lifting weights if you wanna be strong. [00:04:44] Um- [00:04:45] Yeah ... and the biggest challenge that we have right now is that people wanna talk about the promise of the new, but it's much harder for the people who are actually being asked to get it done. [00:04:54] Kiri: Mm. So that, that is a really great entry point here, because [00:05:00] this foundational data work to actually get that done inside of a retailer, who typically owns that? [00:05:08] Why is it so hard for it to get funded? And, you know, we're, we're, we're here talking about retail media. There's, there's [00:05:15] presumably, back to what, what our first, um, one of our first conversations here, that often the, the infrastructure work and the unification work gets lumped with the retail media network leader, even though it's not, like, [00:05:30] ostensibly their job to, to do all of that foundational work. [00:05:34] What really needs to happen? Like, what, what are the kind of the steps that the retailer needs to take to really start thinking about this? [00:05:42] Anne: Great question. So I will refer to a [00:05:45] couple use cases that I know of. I'm gonna not name names, just so that we don't have to worry about clearing this with the respective PR teams. [00:05:52] Mm. I'm thinking of a couple specific examples that I'll refer to. Um, one primarily is the emergence of what is [00:06:00] titled as a chief digital officer- So the biggest challenge that the first retail media networks had, alongside e-commerce by the way, is that they sat in separate departments, and at some point they laddered up to either a CFO [00:06:15] or a COO, CEO. [00:06:16] But at that point, those were really almost sort of board level decisions that were, were being made. What you see with the emergence of this role of chi- chief digital officer, maybe it's hard to say, but it's easy to understand. Mm-hmm. Which is that this [00:06:30] person is responsible for the experience of the site. [00:06:34] This person is responsible for e-commerce. This person is likely responsible for something like this profit center of retail media. Very often it still sits under the chief marketing officer, just to be clear. [00:06:45] Um, but the emergence of the CDO means that that type of infrastructure work that is an input is leading to- Mm [00:06:53] an output, which is cleaner data. And by the way, that is where AI is really going to [00:07:00] shine, is being able to pull... I know we've talked about data lakes for years and years and years, and now what you're seeing is the harmonization of that data. Um, I'll g- I'll actually use an example. So, Miracle has a product called the catalog platform, [00:07:15] and two years ago it won Best Technology at NRF Europe, and it really just goes in and looks at the biggest problem that we had encountered when retailers were trying to launch marketplaces, which was gaps in the catalog data. [00:07:29] Mm. [00:07:30] Again, great, this is not sexy stuff. No. You're gonna go look at- [00:07:34] Kiri: It's not what people wanna hear. It's [00:07:36] Anne: not what pe- people don't wanna... People don't say when they're in third grade, "I wish I could go work on gaps in catalog data and e-commerce." You know, that's like, this was something [00:07:45] that was an emergent issue for our clients, and we said, "Hey, how do we help them make sure that every single product can be shopped when they go and they launch a marketplace, and they surface the catalog?" [00:07:55] Mm. By the way, those kinds of gaps are exactly why you wouldn't [00:08:00] surface in an LLM. And so the fact that we've been solving this problem for literally years now means that the product has sort of met its moment, and that now that we're having these conversations that these retailers are wanting to surface within LLMs, they're going back and [00:08:15] asking the question, "Hey, how do we really get this right?" [00:08:17] So first it's just chief digital officer, um, and again, we, we've sort of mirrored that on our side with just the, the clarity of, um, how the data gets harmonized. And again, I'm, I'm thinking of specific use cases here. [00:08:30] Uh, the second piece that ends up being really important is just a commitment to what success looks like. [00:08:36] Because a lot of what has happened is people were rushing to get the headline, and then if you were to go check in on it, you would say, "Oh, well, how's that going?" And you get the [00:08:45] same answer, "Oh, you know, we haven't really done anything with it." Mm. It's like- Mm-hmm ... okay, well, what was the intended purpose other than just, you know, a bump in the stock or whatever ended up being. [00:08:56] And so the second, I like to say begin with the end in mind. And so it's a lot of [00:09:00] resetting a vision. I mean, I've been having a conversation with a retail media network, um, who's going through a three-year strategic visioning process. The idea that any of us are looking out three years feels scary. Right. [00:09:13] Yeah. [00:09:15] But you also kind of, you have to aim at something if you're gonna swing on the rope to get there. Mm-hmm. [00:09:19] Kiri: I'm [00:09:19] Anne: picturing Tarzan in this, uh, scenario. Um, but, so this idea of, "Hey, are we gonna still stay ambitious about what we think success looks like? And are we gonna do scenario planning? And are we gonna have [00:09:30] a portfolio of strategies for three different outcomes that we think are possible?" [00:09:36] And so it's really just going back to the hygiene of not getting, uh, thrashed by the headlines, and really [00:09:45] going in and asking- Mm ... what matters in the business. And then the third, we had a, a mantra when I was at Clorox about 10 years ago, almost 15 years ago now, which is crazy. Um, we had a mantra, which is we would always ask, "What is our core competency? [00:09:59] [00:10:00] What is the thing that we are uniquely responsible for and that we are uniquely good at?" And you would ask it at the team level. I was a digital center of excellence, is what it was called. Uh, you would ask it at the team level, you would ask it at the division level, and then we as a company were constantly asking, "What is our core [00:10:15] competency?" [00:10:15] Mm. And so I think when you look at retailers who are trying to get- data correct, one of their core competen- competencies has to be to, to commit that digital's a meaningful part of their strategy on a go forward. [00:10:30] Everyone loves to pivot back to that story of, you know, most of transactions still happen in the store. [00:10:35] Great. [00:10:35] Kiri: Right. [00:10:36] Anne: But not forever. And you have to look at- Yeah ... where is the shopper a year from now? Where is the shopper two years from now? And so when you're [00:10:45] thinking about the fact that the center of gravity today might still be brick-and-mortar, you have to ask, "Where's the customer going? Where's the shopper going?" [00:10:53] Kiri: Well, yeah, absolutely. And I, I'm still... I think that there is a great case to [00:11:00] invest in the store, wi- without a doubt. But also, so many people are ta- are, are actually using their phones to supplement their, their shopping experience in the store, and they are using AI chat assistance to [00:11:15] help them research and narrow down. [00:11:17] While they are in the store, they're pulling out- Yeah ... ChatGPT, they are pulling up other AI tools to assess, like do I get, even, even in CPG, these crackers or those crackers if I'm trying [00:11:30] to increase- Right ... my fiber intake. You know, like, so there isn't a, the there's no pure online and pure offline, um, purchase journeys anymore. [00:11:38] They're all interconnected. So what you do online is very important to how, you know, h- how your [00:11:45] brand's brand and the retailer is really showing up in these, um, tools that consumers are increasingly turning to to help them narrow down and decide. That's, that's also kind of a relevant [00:12:00] discussion for y- for the store as well. [00:12:02] Anne: That's right. That's exactly right. And if you start with the customer and you say, "Where is the customer?" One of the answers is she's on the app while she's in the store. Yeah. And so I think that's, like that's an amazing insight. I really love that. [00:12:14] [00:12:16] Kiri: Well, Anne, thank you so much for joining me for this series. [00:12:18] We talked about a few demons, uh, inside retail media. I feel like we have, uh, done a, done a good enough jo- job of trying to coax them out, if not actually exorcize [00:12:30] them forever. Thank you for joining me, Anne. I'll, I look forward to seeing you at, at Cannes at our breakfast. [00:12:36] Anne: That's right. We're having breakfast together at Cannes. [00:12:38] Uh, it's the Thursday of that week. We're welcoming retail media leaders, retailers, [00:12:45] people who are representing the hardest part of the business, which is really getting it right and making sure that we're compelling shoppers along their journey. [00:12:52] Kiri: See you there [00:12:53]