Subscribe
Share
Share
Embed
10 minutes of expert insights every weekday. Your morning ritual for staying ahead in retail media.
Kiri editorializing
===
[00:00:00] Kiri: Andrew Cardo keeps seeing the same thing happen with retail media networks. A brand asks for data that they can feed into their marketing mix model and but the retailer misunderstands the request and tries to sell them on an attribution [00:00:15] dashboard instead, and the brand walks away.
[00:00:18] Andrew has spent over 15 years building measurement programs at Google Meta, Netflix, and Snap. And he now runs Growth by Science, a consultancy, helping [00:00:30] brands build custom incrementality testing and media mix models. So he's seen this from every angle. And on a recent episode of the Unlocking Retail Media podcast with host James Avery, he he laid out [00:00:45] why Ironmans keep losing credibility with measurement. It's not that the ads don't work, it's that they can't prove it in the language that their most valuable advertisers speak.
[00:00:57] This is a great episode. I learned a ton [00:01:00] and I'm gonna share three things that I took away from this conversation about what retail media networks are getting wrong on measurement. Let's jump in.
[00:01:08]
[00:01:09] Kiri: Number one, the big ad platforms stayed measurement agnostic, but retail [00:01:15] media didn't.
[00:01:16] Andrew says that platforms like Meta, Google and Snap got big in part by supporting whatever measurement approach that advertiser wanted to bring. Incrementality, MTA media mixed modeling, if [00:01:30] there was demand, they built for it, but RMNs tend to do the opposite. Let's listen.
[00:01:35] Andrew: Of course, having worked at a bunch of them, I, I, I do have a special place in my heart for the ad platforms, but ad platforms are not necessarily in the business of [00:01:45] optimizing advertiser outcomes. They're in the interest of, or they're in the business of a, of optimizing their own revenue. And the way that you do that is you have to be effective.
[00:01:53] And so you can't not optimize for advertiser outcomes, and you can't not be a performing platform, but you [00:02:00] also are always trying to find an edge, um, to, you know, to showcase a story, to showcase the performance of your ads and, and to, you know, create a narrative that attracts advertisers from a different vertical, from various different verticals to your platform.
[00:02:13] And, and really the [00:02:15] way to do that is with measurement. Um. And so the smart platforms will always take a, a relatively measurement agnostic perspective and, you know, kind of, uh, forge partnerships and, and build, uh, first party products, [00:02:30] uh, that are, that are really born of the demand, uh, that they're receiving from the advertisers.
[00:02:35] So if Nav advertisers says, Hey, I really care about. Incrementality or I really care about, uh, MTA or I really care about some random other type of measurement [00:02:45] approach. If there's enough demand for that measurement approach, the advertising platform is gonna support it. And that's why when you, when you do work with some of the bigger platforms, you know pretty much any measurement approach that you want to bring to the table and you want to utilize that will be somehow [00:03:00] supported.
[00:03:00] Now, they may try to guide. You know, uh, advertisers a certain way and, and they may try to, um, you know, talk, talk a lot about some of the more sophisticated and advanced, uh, measurement approaches, but they're, they're never, never gonna turn down business. And if, you know, a [00:03:15] certain measurement approach they disagree with, but drives a lot of business, they're gonna support it.
[00:03:19] So, so really, I think, you know, as an advertiser. You, you know, it's really important that, that you have your own opinions on measurement and that you, you know, you kind [00:03:30] of lean on on your own research and, and your own analytical, uh, expertise to, to kind of set what you want your measurement approach to look like, rather than, you know, take that input, uh, necessarily from the platform.
[00:03:42] You can of course, listen to what they have to say, [00:03:45] but just know that. You know, their, their measurement programs are, are not 100% altruistic.
[00:03:51]
[00:03:52] Kiri: The lesson from Big Ad Tech is simple. Don't pick the measurement approach for the advertiser. Support what they need. [00:04:00] Takeaway number two, retail media networks are answering questions that their biggest advertisers have stopped asking. Andrew says that retailers are over-investing in things like granular user level path to purchase [00:04:15] data while their most valuable prospects actually need something else.
[00:04:18] They need aggregated feeds for modeling and experimentation.
[00:04:23] Andrew: you know, where I have seen measurement lag somewhat, uh, from the ad platform side is [00:04:30] definitely in the RMN space.
[00:04:31] And, and it, you know, it's just born of the fact you, you know, rms. Are, are not typically the primary, uh, revenue stream for, for folks, right? This is, this is, this is retailers or, or other platforms that have a [00:04:45] ton of data and realized, you know, in the last few years that hey, there's some great monetization potential we can help our, you know, uh, various constituents.
[00:04:53] We can provide a great experience to to consumers, but also. You know, help our advertisers and, and, [00:05:00] you know, uh, the wholesalers to, to really promote their products in a good way. And, and so, so they're, by nature, not caught up with, you know, what I would say as natives to ad tech. Um, and so what I, what I tend to see [00:05:15] happen a lot in RMNs is this over focus on.
[00:05:19] You know, really granular user level data, trying to design this path to purchase data set, uh, with the idea that hey, we can, you know, prove that there was a touch point on some [00:05:30] property that proceeded a conversion and that's gonna be good enough, uh, to prove, uh, to prove efficacy. Um, and I would say that that is not the case, uh, at all.
[00:05:39] Uh, that's, that's kind of the old school way of doing things. Um, certainly there's demand for that. You know, [00:05:45] a lot of advertisers still care about those types of data sets, but, uh, the approaches that the most sophisticated advertisers are taking are, are really more in the modeling, the experimentation and, and you know, the, in sort of like that aggregate data, uh, approach.
[00:05:59] [00:06:00] So, um, you know, I think, I think retail media needs to, needs to consider what does. A, an experiment require, and what does an MMM require? And not to say give up on the attribution. 'cause as I said, there's a [00:06:15] lot of demand still for that. But supplement that with, you know, a, a great API that lets you get data out in a, in a way that's conducive, uh, to, to MMM. [00:06:30] 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. Both [00:06:45] advertiser audiences demand a seamless advertising journey from onboarding to reporting.
[00:06:50] Kiri Masters: You can offer everything from sponsored products to video ads all in one solution. Learn more@miracle.com. That's [00:07:00] M-I-R-A-K l.com.
[00:07:02]
[00:07:03] Kiri: And highlight number three. First party lift is grading your own homework. Just be honest about it. James asked Andrew how he designed [00:07:15] incrementality for a retailer, and Andrew laid out a graduation framework. Don't force full incrementality on new advertisers, but also don't pretend basic attribution is the ceiling either.
[00:07:28] Let's listen.
[00:07:28] Andrew: I, I think there's a [00:07:30] couple of approaches, right? You nailed it where there's, there's the, the segmentation of the advertisers. For one, I wouldn't force incrementality on people who are just trying out the platform.
[00:07:39] What I would, what I would do is acknowledge that. Whatever attribution [00:07:45] measurement you might be doing at that level is not gonna be perfect. But, you know, it'll, it can be, you know, sort of directional and it can give a hint as to our people that are buying your products, seeing the ad, right? I mean, that's, that's the, the [00:08:00] most conservative way of, uh, of, of representing that.
[00:08:02] And you can say like, in the best case, everybody that saw the ad bought the product, in the worst case, you're at least showing it to people that would've already bought that ad anyway. And so you know that your target audience is here. Uh, right. And then the next [00:08:15] step after we prove that out is to, is to tweak the incrementality.
[00:08:17] But you know, you've gotta have a certain minimums in, in terms of spend to be able to do that.
[00:08:21] but yeah, first party lift is, is always good. It's, it doesn't cost, you know, obviously the development costs, but there's no, you know, having to partner with somebody else. However, it is [00:08:30] a bit of grading your own homework. I would look at first party lift as, as kind of a more of a sales tool versus a measurement tool and, you know, go into it fully acknowledging like, look, we are doing our best to be objective, but you know, we understand that you may not, you know.
[00:08:44] [00:08:45] Look at this as, as like purely objective. 'cause it isn't 'cause it's our own platform, blah, blah, blah. Um, however, you know, as a graduation from attribution, I think this is a great next step.
[00:08:54]
[00:08:55] Kiri: So the path is this directional attribution for newer, [00:09:00] less sophisticated. Advertisers first party lift as a step up from that, and then third party geo testing and MMM integration for the biggest spenders. A lot of retailers are stuck at [00:09:15] step one, but here is the slightly uncomfortable part.
[00:09:19] Andrew was describing what sophisticated brands need, but a huge chunk of brands aren't even there yet. I spoke with Liz Roche, who is VP of Media [00:09:30] and Measurement at Albertson's Media Collective a couple of weeks ago about their recent eye roll as research, and she made the point that many brand partners receive one of these kinds of.
[00:09:41] Of reports and simply don't have the team to really [00:09:45] question or understand what's behind it. Not every brand has a data science team that can evaluate whether that IR OAS figure was calculated using propensity score matching or clustering or whatever. Historical brand [00:10:00] sales were included as a feature.
[00:10:02] That research from Albertsons found that 83%. Of campaigns can flip from positive to negative eye OAS based on the methodology alone, and this means that transparency [00:10:15] isn't just a nice to have for the top tier CPGs running mms. It's also essential for the mid-market brands who are taking the number at face value.
[00:10:24] So the takeaway here is that retail media has a measurement gap at both ends. The [00:10:30] most sophisticated brands can't get the data formats that they need, and the mid-market brands can't evaluate what they're getting. Andrew's advice, RMNs is the same, regardless. Be honest about what your measurement does.
[00:10:44] And doesn't [00:10:45] prove and build the APIs and geotargeting that advanced advertisers require. Invest in internal expertise that really understands the why behind the ask. Check out the Unlocking Retail Media podcast for [00:11:00] the full conversation. Thanks for listening, and I'll catch you tomorrow.
[00:11:04]