Retail Media Breakfast Club

What if the biggest opportunity for advertising in AI isn't just converting shoppers who are ready to buy, but reaching them much earlier in their decision-making journey? Today I'm re-sharing the newest episode from my other podcast, AdGentic, where I was joined by Roger Dunn, Chief Commercial Officer at Thrad, to unpack why brands need to rethink how they approach advertising in AI environments.

Roger brings a fascinating perspective from his 20 years in retail media, and together we explore how advertising in AI is evolving beyond traditional search and sponsored product placements. We discuss the full-funnel opportunity for brands, the emerging formats for AI advertising, the role of retailer and personal agents, and some important questions around brand-controlled information, consumer trust, and where this rapidly evolving channel is headed.

This episode is sponsored by GrowthLoop

Timeline

[00:00] - Introducing Roger Dunn, and exploring why AI is opening up an entirely new frontier for retail media and advertising.
[02:38] - What advertisers get wrong about advertising in AI environments
[04:24] - Show and Tell: Why AI advertising opportunities span the full funnel
[06:33] - Why upper-funnel AI advertising could deliver greater incrementality
[08:00] - Could AI conversations shift retail media investment beyond sponsored products?
[09:33] - Who does an AI agent really work for: the shopper, retailer, or brand?
[11:00] - Why brand-controlled information and AI accuracy matter more than ever
[14:28] - What's next for sponsored agents and the evolution of AI ad formats

Links & Resources

What is Retail Media Breakfast Club?

10 minutes of expert insights every weekday. Your morning ritual for staying ahead in retail media.

Ads In AI Can Span The Full Funnel, Says Thrad's Roger Dunn
===

[00:00:00] Kiri Masters: Good morning, and welcome to Retail Media Breakfast Club. I'm your host, Kiri Masters, and in case you missed it, I have a new newsletter and podcast called Adgentic. It's all about the emerging ecosystem of [00:00:15] ads in AI interfaces, and that includes the regulars like ChatGPT and Perplexity,

[00:00:21] But also AI surfaces inside of retailer apps. AdJantic is part of the Refi by Media Network, and you can subscribe to it [00:00:30] wherever you're listening to this podcast. Today, I'm sharing an interview that was published last week, reproduced with permission, originally titled Ads in AI Aren't Just for the Bottom of the Funnel

[00:00:44] Let's [00:00:45] jump in

[00:00:45] Roger Dunn: It depends whose side that agent is working on. The personal agent is on the shopper's side supposedly, and then the retailer's agent is on the merchant side[00:01:00]

[00:01:02] Kiri Masters: This is Adgentic, the series about advertising in the agentic era. I'm Kiri Masters, someone who has spent the last 10 years analyzing the retail media ecosystem, but [00:01:15] now I'm captivated by this whole new era of advertising, ads in AI. Today we have a special guest, but we like to do things a little bit differently around here at Adgentic.

[00:01:26] So this isn't your standard how's your uncle interview. [00:01:30] Every guest who'll join us brings visuals to show and tell. So if you're just listening via audio, we'll do our best to explain, but you might also like to check out the companion YouTube video or written essay, all [00:01:45] available at adgentic.ai. In this episode, I'm speaking with Roger Dunn, chief commercial officer at Thrad.

[00:01:54] Thrad describes itself as advertising infrastructure for conversational [00:02:00] AI. Brands buy ads inside of AI apps, and those apps get paid to show them. Thrad says its network reaches over a billion users a month across hundreds of AI publishers in more than 180 [00:02:15] countries. Before Thrad, Roger spent 20 years in retail media, most recently leading global retail media at Diageo.

[00:02:23] He also writes the AI Commerce Brief on LinkedIn, an excellent newsletter we'll [00:02:30] link up to in the show notes.

[00:02:32] Roger Dunn: Oh, and at Adgentic, we also like to get right to the good stuff, so let's jump in.

[00:02:38] Kiri Masters 2: Now, in your capacity as chief commercial officer at Thrad, what do [00:02:45] advertisers, you know, brand advertisers specifically, get wrong about ads in AI environments right now?

[00:02:52] Roger Dunn: Well, I think there's a, there's an obvious comparison to things like search. Um- Mm ... and that's not an unfair comparison in the sense people are [00:03:00] entering in something, some intent- Mm-hmm ... and then they're, you know, realizing, you know, having something given back to them in terms of content coming back. So there is a fair comparison, but it does operate a little bit differently, obviously, as we know.

[00:03:12] A typical keyword search is maybe only two or three [00:03:15] words, whereas a typical prompt is, you know, 150 words, say. So the, the, the level of intent, the level of kind of context that's provided by, um, the user when they, on the input side of things is, is very different.

[00:03:28] Kiri Masters 2: Then so are you sort of funneling [00:03:30] demand as well from your advertiser base into all of those different chat environments?

[00:03:36] Roger Dunn: Yeah, so the opportunity for the advertiser is to be, to have a much more diversified, uh, media plan essentially. So accessing [00:03:45] conversations wherever they're happening, so it's not necessarily... It's about the context of the conversation, of course. So we're targeting like prompts, personas, keywords even within that environment.

[00:03:56] Um, but it's where can we find that particular conversation? And [00:04:00] then obviously we're trying to make sure that the ad experience and working with the publishers is the right one. So we might not have an ad appear on the first turn of conversation. It might, there might be some parameters there where it comes in later at the right time when it's relevant and n- at the [00:04:15] right frequency as well.

[00:04:15] So managing that is an important part of it, and we want it to be a sustainable, um, channel, um, both for advertisers and the, the publishers themselves.

[00:04:24] Kiri Masters 2: And now it's time for my favorite part of the show, Show and Tell. Show and [00:04:30] Tell Every guest on Agentic brings visuals, a chart, a diagram, even a sketch on a whiteboard.

[00:04:38] It might be from their own research, a slide from their pitch deck, or a framework for how they see the [00:04:45] market. Then we use it to steer the conversation. For his first show and tell, Roger brought a funnel diagram, and it takes one example advertiser, Netflix, and maps nine [00:05:00] AI conversations where an ad could win a new subscriber.

[00:05:05] At the top of the funnel, Netflix isn't even mentioned. Someone types something like, "Date night in. What should we watch? Funny, under two hours." The AI builds [00:05:15] the list, and in the middle of the funnel we see rivals sitting side by side, a prompt like, "I get Prime Video with Amazon. Do I still need Netflix?"

[00:05:24] And at the bottom of the funnel, agents start acting. "Cancel any service we haven't [00:05:30] used in 60 days. Keep Netflix." Now, all of these prompts are illustrative, not real search data, but what I like about this diagram is that it challenges a lot of assumptions that advertisers have, that ads in [00:05:45] AI belong only at the bottom of the funnel, converting someone mid-conversation.

[00:05:50] But Roger is showing that people turn to AI right at the start of their research too, and that is a real chance to reach them. Now, I'd argue that the ad [00:06:00] formats still need a little bit of work before they deliver a really great discovery experience, but there is still a case for an ad at every stage and not just at the point of purchase.

[00:06:12] Roger Dunn: And obviously this is just an example, so we could easily have [00:06:15] had, like, pet food in here as like a CPG- Yeah ... example where people are doing a lot of discovery around pets and, like, nutrition and, and then getting down into products, then getting down to purchases. So yeah, it is just an example, but it, it does show that full funnel and, like, [00:06:30] the different, different stages that people will move through.

[00:06:33] Um, and obviously as we know with that lower funnel thing, uh, uh, ad opportunities rather, um- They're converting people potentially that were already gonna buy anyway, and you [00:06:45] might actually move the dial more on incrementality, which we know is an important thing in retail media, but in any advertising further up that funnel.

[00:06:53] Because people are yet to decide, they're in that moment of decision, and actually that's the moment to win people. Um, [00:07:00] but you know, obviously opportunities throughout. Are manual processes slowing down your RMN momentum? With GrowthLoop's [00:07:15] composable commerce media solution, retail media and ad teams can build and launch precise audiences in minutes, activate across DSPs and ad platforms, and deliver transparent [00:07:30] ROI reporting that drives advertiser revenue.

[00:07:34] Kiri Masters: learn how retail media leaders at Costco, Fanatics, and goPuff use GrowthLoop to deliver stronger results for their brand partners [00:07:45] fast. Visit go.growthloop.com/breakfast. That's go.growthloop.com/breakfast

[00:07:58] Kiri Masters 2: Now, that last point [00:08:00] should make retail media sit right up in their chairs, because sponsored products live at the bottom of the funnel right at the moment of purchase.

[00:08:09] And if AI conversations become the place where people decide, a lot of that money has a [00:08:15] reason to move up the funnel. Which brings me to Roger's next visual, which breaks down the agent ad formats that are already out there. He sorts the options by the following question: Who's doing the talking?

[00:08:29] [00:08:30] Sometimes you click an ad on YouTube or on ChatGPT and chat with an agent for that brand. Sometimes the ad opens a chat in your messages on WhatsApp, on Snapchat, or TikTok, and sometimes the platform's own assistant [00:08:45] does the talking and the brand pays for it, like Alexa on Amazon or Sparky at Walmart.

[00:08:51] Roger Dunn: This is sort of like the digital version

[00:08:54] Kiri Masters 2: of I'm in market for a new car, and I go to the different [00:09:00] showrooms and I speak with someone on the floor about, "Hey, I'm looking at this model," and they're sort of helping me figure out what type of car I might need. It's consult- it's consultative. It's, there's some back and forth.

[00:09:12] I know that I'm talking with [00:09:15] Mazda, right? It's like a Mazda sales agent, um, where it's, it, it's clear that this is a sales conversation. I'm open to it. Is that a ... How does that [00:09:30] comparison sort of stand up in your view?

[00:09:33] Roger Dunn: Yeah, I think that's the, the interesting to think of, like, whose perspective the agent is coming from, right?

[00:09:38] 'Cause if you're, if you're on Amazon or Walmart, you know that agent is, is obviously owned by the platform you're on, by the [00:09:45] retailer. So they're, they're gonna be, um, obviously trying to give you all the facts and figures, but they're, they're, they're there to sell you product as well. So and I think people are very aware of that, and I think that's fine.

[00:09:56] Whereas, whereas if you've got these personal agents which are coming from [00:10:00] outside or maybe general purpose ones like ChatGPT, I guess they don't have a particular merchant in mind. They don't even have a particular product in mind. In theory, there, this neutrality is, is, is more central there. Um- Mm ... and obviously, yeah, it's, [00:10:15] it's depends whose, whose side that agent is working on.

[00:10:18] The personal agent is on the shopper's side supposedly, and then the retailer's agent is on the merchant's side, so. Um, and then obviously we haven't mentioned brands here, but the, the brands are coming through in, in the Google and [00:10:30] YouTube example where it's actually based on the brand's information and content.

[00:10:33] So, um, although it's plat- platform managed by Google, it's, it's the, the brand's kind of perspective being brought into the conversation.

[00:10:42] Kiri Masters 2: Yeah. That, that's, that's what my assumption [00:10:45] was with the ChatGPT-sponsored agent, is that the, uh, the agent is, you know, it's being, it's being run by ChatGPT, but it's pulling the information that the brand- Mm

[00:10:56] has provided to it, and that is also at [00:11:00] Amazon, um, Unboxed last week, something that they announced there as well, right? That, um, their sponsored agent, essentially that brands are now going to be able to provide Alexa with the corpus of [00:11:15] brand-provided information, because I think the concern that a lot of brands shared back with Amazon, and something that I pointed out when I was l- looking at it, at, at Alexa is, what about incorrect [00:11:30] information that other people-

[00:11:31] perhaps consumers or maybe competitors have put out there about our brand where it's incorrect, and, and, uh, Alexa is assuming that, you know, something that people have been saying about, uh, the item [00:11:45] that's not correct, um, that actually makes it into the AI answer. That's a huge concern for brands- Mm

[00:11:51] teams. Or if, you know, they change their packaging or they change their formulation or something like that, they need to make sure [00:12:00] that the agents are out there providing the most current and up-to-date information, not just what was sort of, uh, captured, uh- Mm ... in the, in the past.

[00:12:11] Roger Dunn: Yeah, it's ve- it's very true, and it's, in some cases, the information that is being [00:12:15] used in these agentic experiences was never intended for that.

[00:12:18] It was a product feed, or it was a merchant center update that wasn't intended to- it was intended to be 100% accurate, but not necessarily for that use case and that purpose. Mm. There was also an [00:12:30] interesting update from Anthropic, who launched some commerce agents for retailers. Um, and they were saying in some of their best practice that it should take all of the brand information and product information obviously as [00:12:45] 100% accurate, but when it's reviewing and synthesizing things like reviews, which are actually UGC, then you need to have greater guardrails around that, and like- Mm

[00:12:54] not take instruction from necessarily that UGC content, but take instruction only from [00:13:00] brand owned and operated content. So I think there's definitely some, some nuance to how you kinda set this up, and like how you think about it from a, a brand perspective or a retailer perspective for that matter.

[00:13:11] Kiri Masters 2: Yeah. Yeah, it's, it, it is very interesting [00:13:15] because as a, as a consumer making a decision, I do wanna hear what other people who've used that product have to say. I want legitimate user feedback. But we both know that those feedback e- [00:13:30] ecosystems are often gamed. And, uh, you know, gamed at, at worst, but at best, not everyone jumps on Reddit to share their experience or leaves [00:13:45] reviews for anything really, booking hotels and airlines and expensive cars and things like that, that very few of us actually follow through and provide- Mm

[00:13:55] a, um, completely honest review on- Yeah ... a- anywhere [00:14:00] that these, um, systems are scraping that content. It seems like a new type of sponsored format that we haven't seen anywhere else in the digital advertising space. [00:14:15] Do you think that this is like the, the play, or are we at version 1.2 of, you know, an infinite number of different ad types in AI services?

[00:14:28] Roger Dunn: You have to imagine the [00:14:30] ads are gonna evolve from here. Like even as we've mentioned already, like in the last few weeks we've had these sponsored agents come on board. ChatGPT obviously just launched this visual ad as you're doing an image generation. There's this moment in time as you're waiting for that image to generate [00:14:45] where an ad is, is gonna be shown.

[00:14:47] So yeah, y- obviously there, there's gonna be like an evolution of th- of this. But I think- Whereas where the opportunity is, obviously, as people are entering in this natural language, whether [00:15:00] they're... And they might be speaking to it more, so even longer conversations, the intent and the context that is being provided in that is so much greater than someone just browsing over a banner or entering two keywords in to find some trainers, say, on, [00:15:15] on Google.

[00:15:15] Like, the, it's, it's the, the wealth of, of that ta- the ability to target the wealth of that information that I think is the real, um, game changer here. Mm. Um, and then there's just the things... I guess the other perspective is, like, what, [00:15:30] what do these platforms know about you? So a retailer's agent might be very verticalized.

[00:15:35] It knows all your shopping history and all of the details of, of what you've shopped on there before, what you've been browsing in the moment on that retailer. But if you think about those broad-based [00:15:45] LLMs, they have a much more broad view on you and have a history that is much broader than just your shopping even.

[00:15:50] They know what else you're kind of thinking about and asking questions about. Mm. So that broader perspective on the consumer, and then the personal agent that sort of evolves out of that [00:16:00] is gonna have a very different view on the consumer, and obviously gonna be help- be able to help out in different ways, and therefore will realize different commercial opportunities than, than a retailer.

[00:16:11] Kiri Masters 2: Mm. Fascinating. Well, is there anything else [00:16:15] that you want to leave the Adgentic audience with today, Roger?

[00:16:19] Roger Dunn: Yeah, great to be here, Kiri. And, uh, yeah, people, if they want to kinda get in touch, LinkedIn's probably the best place, so connect with me there, and then I've got the, the AI Commerce Brief, which is my weekly newsletter, [00:16:30] if people are interested in this space.

[00:16:31] I'm gonna try and keep up on everything. It's, um, bit of a, bit of a challenge with the amount of announcements- Oh, yeah. ... um, that are coming out. But, um, yeah, feel free to follow along. And then, yeah, obviously it could be great to hear from people if they've got some experiences in this [00:16:45] space or if they're interested in the, the kind of ads in LLM space.

[00:16:49] Um, certainly a, a new channel, a new opportunity for many advertisers and agen- agencies.

[00:16:54] Kiri Masters 2: And that's Roger Dunn from Thrad. Thanks for tuning into Adgentic, the show [00:17:00] about advertising in consumer AI environments. Adgentic is part of the Refy By Media Network. I'm Kiri Masters, and I'll see you next [00:17:15] time.