← The Marketing Architects
The Marketing Architects
The Science of Ad Personalization
00:25:53
Personalized ads outperform generic ones. But the effect is smaller than most marketers expect, and the hidden costs can quietly undercut your brand. This episode, Elena, Angela, and Rob dig into a meta-analysis of 53 studies on ad personalization...
Loading summary
Transcript57 lines
- [00:00]
A
Honestly, the biggest pain point for us in post production is just the time that it takes to put these things together. You really have to sacrifice quality and creativity just to kind of wrap your head around how many spots we're actually trying to put out, putting them out, accurately, matching up Iski codes. It's just a massive endeavor to try and put out something at scale, even just a couple hundred spots.
- [00:22]
B
Marketing Architects hello and welcome to the Marketing Architects, a research first podcast dedicated to answering your toughest marketing questions. I'm Lena Jasper. I run the marketing team here at Marketing Architects, and I'm joined by my co hosts, Angela Voss, the CEO of Marketing Architects, and Rob DeMars, the chief product architect at Misfits and Machines.
- [00:42]
C
Hello.
- [00:43]
B
Hello. And we're joined by two guests today. Matt Holtgren, our Chief Analytics Officer at Marketing Architects, and Josh Wilson, the director of AI Audio at Misfits and Machines.
- [00:53]
D
Thanks for having me.
- [00:55]
A
Hey, y'.
- [00:55]
C
All.
- [00:55]
A
Happy to be here.
- [00:57]
E
Welcome, Matt and Josh.
- [00:58]
B
We're back with our thoughts on some recent marketing news. Always trying to root our opinions in data research and what drives business results. Before we jump in today, I wanted to mention that we are going to be at Digiday's Modern Retail Marketing Summit at the end of the month. We'll be speaking and we'll have a fully stocked popcorn booth. So if you're going to that event, we would love to meet you. But today we're talking about ad customization. What does the research say about ad personalization? When does it work? When does it not? And how might you approach personalization? When you are in a mass reach channel like television? But I'm going to kick us off, as I always do, with some research. And for this episode I chose a meta analysis. This was published in the Journal of Advertising Research and it's titled How Persuasive Is Personalized Advertising? A Meta Analytic Review of Personalized Advertising. With all the data available today, marketers can tailor ads to individuals based on their behaviors, their interests, their demographics. And if an ad feels more relevant to us, we would think it should be more persuasive. But how much does personalization actually help? This study analyzed 53 experimental studies with nearly 12,000 participants comparing personalized ads to generic ones. And what they found was that personalized ads do outperform non personalized ads, leading to slightly more positive attitudes towards the ad and stronger behavioral intentions. But the effect size was relatively small. What really drove the impact was perceived relevance. So that's when consumers felt like the ad related to them personally. Then they were more likely to engage with it. And what's interesting is the study found that personalization didn't significantly increase feelings of intrusiveness, suggesting people often appreciate relevance more than worry about being targeted. So personalization can make advertising more persuasive, but it appears to be a modest effect, which raises a larger question for marketers. How far should we go in customizing our advertising? And first, I'm curious how we would answer that question ourselves, because in the past on this show we have been wary of certain kinds of targeting. And personalization generally has to come with some layer of targeting when you're looking for a more specific person or audience group. Ang, what are some of the marketing effectiveness watch outs that come with ad personalization?
- [03:05]
E
Yeah, I'd say the biggest one is just the reach trap. Right. So when you personalize, you're almost always narrowing in terms of audience and shrinking that audience has real costs. You lose that mental availability that comes from reaching people who aren't actively in market yet. We know brands grow by recruiting new buyers and they can't do that if they're only talking to existing customers or type behavioral segments, things like that. I think there's also a measurement trap. Personalized campaigns also often look great from an efficiency standpoint. Better ctr, better conversion rate. But when you're reaching a pre filtered audience who was possibly going to convert anyway, that's not the same as being persuasive. And it can be a flattering lie that the data tells you. And then I think the third to watch out for would just be brand consistency, which we can get into more later. But when you start versioning creative for dozens of segments, you can quietly erode what makes your brand distinctive. Every version is a decision about what to say and what to cut out. And those decisions can just compound over time.
- [04:11]
B
Yeah, you definitely have to be intentional about distinctive assets and making sure it's consistent. One thing I was wondering is sometimes when we talk about personalization, I think it comes from maybe a misunderstanding or an overestimation of my audience and how I think that different audience groups really need different messaging. They're very different from each other. And do you think sometimes marketers overestimate that? Like how much different slices of their audience might need to different messaging?
- [04:38]
E
Absolutely. I think it's one of the most common and costly mistakes that we see. The assumption is that different people have different needs and therefore they need a different message. But to your point, the research keeps showing that the things that make advertising work, emotion, memory, distinctiveness salience work on humans very broadly. So a funny ad is funny to most people. A brand story that builds mental shortcuts does that for almost everyone who sees it. Where marketers get tripped up is confusing using, I think that cultural relevance with audience segmentation. So making sure your creative doesn't feel tone deaf to a particular audience is different from building 30 versions of an ad targeting behavioral micro segments. The first is just good creative, the second is expensive and often unnecessary. And I think there are real cases where message differentiation matters. Say different messages for perhaps people in different phases of the purchase funnel, people who have never heard of you versus people who are very familiar. But for brand advertising at the top of the funnel, most of the audience wants roughly the same thing just to understand what the brand stands for and why it matters to their life.
- [05:53]
B
Yeah, I was going to say definitely matters what stage of the funnel you're in. And we saw on LinkedIn the other day, I think both of us were tagged in this post an example of a B2B billboard that was so niche and in an area like out of home, it was almost so complex, like only one decision maker could understand it and would they even understand it at first glance. So I think some of that too is the broadness of the channel you're using. And just also people are not paying a ton of attention to your ads. So if you can be more simple sometimes and more broad, that can be better as well in an upper funnel channel. Matt, we've talked in the past, as I mentioned earlier, a lot about targeting on this show and that we should be wary of targeting that comes with a high cost. So I'm curious, what kind of targeting methods have you seen struggle that you would say to a brand or a marketer, you should avoid these kinds of targeting when you're personalizing an advertisement?
- [06:44]
D
I think Ange started to steal the words out of my mouth, but the first place my head goes is hyper targeting those small segments. And it really has less to do about the personalization and more about the fact that that approach kills your reach. And reach is still doing most of the heavy lifting when it comes to channels like LINEAR and ztv. Now it's really easy to sell fancy third party list and they sound better than unfortunately what they actually are. You have to realize the channel, what you're working with, like CTV is still predominantly targeted off of IP matching and that's genuinely hard. I live and breathe it every single day. So that precise behavioral audience you're really paying that premium for just isn't nearly as targeted as it looks and sounds on paper. And I think the research backs that up too. That what drives persuasion isn't whether or not you're targeted, targeted someone correctly behind the scenes, it's whether the person actually feels that is relevant. So if you're burning budget on expensive lists that just shrink your reach, you're paying more for less.
- [07:44]
B
Yeah, that was one interesting part of the study that they found that the effects of personalization, they did find positives, but it was tough to get to that. It seemed, I would think as a brand, if you're testing personalization, you'd want to make sure it's working. So as a measurement expert, how, if you were looking at campaign results, how would you separate the impact from personalization versus everything else that's happening in a campaign to determine if the sometimes added cost is worth it?
- [08:11]
D
These are lessons you can take beyond personalization. And you should think of every test this way. Because I always put it this way. I hate tests that land in the gray zone. Like, you need to set up your test in a way that you're going to know either either worked or didn't work. And that really comes from being strategic up front. Use things like geography to your advantage. So when you're running the customized version, have your A versus your B, obviously your test markets get exposed, half of it gets exposed the control, half of it get exposed the test. And from there you can get a cleaner read. If you don't do that and you just go blast things out nationally or think about other variables. The number of times where clients test three things at once and they're like, oh, my thing won. No, my thing won. And teams are now fighting for what actually caused it. It's mind numbing. So you really need to do the due diligence up front. Plan it clean and make sure that you've set it up. The other thing too is defining what's worth it or what you're measuring up front too. I think sometimes people are so giddy just to test, they think of that as an afterthought. And they're like, oh, wait, I didn't even set this test up in a way to try to achieve the thing I was trying to achieve. My basketball coach in college always said, slow is fast. Do the planning, make sure you got it right so that you can actually get the outcome you want.
- [09:25]
B
Yeah, that's advice I should apply to my personal life as well. I think that's just general good life advice too.
- [09:31]
C
Did you ever tell him he's wrong, though.
- [09:34]
D
That was not a good idea.
- [09:36]
B
I saw a funny video yesterday that said, there'll be nothing harder in life than your college sport. I did not play a college sport. But I'm like, man, you hear that sentiment a lot from people that were college athletes. I know Matt and Ange were both college athletes, but. Rob, question for you. Ange mentioned in the beginning, one of the struggles or watch outs when you're personalizing advertising is how do we keep our distinctive assets consistent. And that can be hard when you're creating a lot of different versions. Do you think that's a risk of personalization, that you're going to weaken your brand consistency over time?
- [10:10]
C
Ange mentioned a lot earlier, she's so thorough and thoughtful. I'm not even sure why I show up. Another way to put it, she sort of stole all the thunder.
- [10:19]
E
I did not. You always got good stuff to say, Rob.
- [10:22]
C
I'm kidding. She hit on all the right points. And I do think just to continue down that road, it really only weakens the brand customization and personalization when you confuse personalization with changing your identity. And I think that's what really comes down to customization should only change potentially the arguments that we're making to the customer based on their needs. Maybe a CFO sees a message about roi. The end user of the product might see ease of use, but the colors, the typography, the sonic logo, the core visual architecture, you know, perhaps the overall campaign idea that should have the legs to be able to transcend all of those potential personalizations. Those things just, they need to be locked in the vault. The moment a creative director, you know, they're like, I'm going to tweak the brand color just to make this version pop a little more. For whatever personalization reason they're doing. That's when it's really, you're not personalizing the ad anymore. You're actively destroying the mental availability of the ad.
- [11:27]
B
So that's definitely a challenge with personalization is how do we maintain distinctiveness. But there are other issues too. If you think about creative side of personalizing, there's the cost of it. The danger if you get it wrong. I'll say from my point of view. We got a direct mail piece the other day that was really cool. It was about, I think, some sort of grass product, but it had a picture of our house on the mail piece, which is really cool. But then it got our name wrong and I was like, dang, that was almost like such a good marketing piece. But it said our last name was James, not Jasper. And that can happen. We've all experienced this at a worst level on a channel like digital. So what do you think is the biggest creative roadblock to. To personalization?
- [12:11]
C
Yeah, I mean, all the things that you listed are really important, but to me, TV is near and dear to my heart. So I'm going to think about it for a minute through that lens. Personalization in TV is difficult because of just the mere supply chain nature of production. Historically, if you wanted 5,000 versions of a broadcast spot, first of all, you probably would have never even gone there because that seems impossible. But if you want a lot of versions that, you know really comes down to the unit economics, you could completely bankrupt your budget trying to create that kind of version. And then not only is it too expensive, but it's just slow. And part of the benefit you can sometimes get from personalization is timeliness. Right. And you really lose that traction when you have to spend weeks creating all those versions. But we're at a super interesting crossroads, I think, for television and technology where we're able to really reimagine what's possible in production. Because of tools like AI and being able to do personalization, TV is no longer becoming that expensive and time consuming. You just really need to reimagine the process, build a better machine.
- [13:26]
B
Yep. Which, spoiler alert. As a TV agency, we believe we've done this. So instead of just avoiding personalization altogether, we've been thinking about how could we do it well in tv. So part of the inspiration for this episode and diving into customization in general is that we've built something that we believe can bring customization to a mass reach channel like TV while maintaining the marketing effectiveness principles we all care about. It's called the mass customizer, and we have the person who built it on the show today. Josh, thanks again for joining us. I know Rob just outlined some of the challenges with creative production and personalization, but what in your mind, in your opinion, are some of the traditional roadblocks or challenges with customizing something like a TV commercial at scale?
- [14:09]
A
Yeah. So honestly, the. The biggest pain point for us is in post production is just the time that it takes to put these things together. You really have to sacrifice quality and creativity just to kind of wrap your head around how many spots we're actually trying to put out, putting them out, accurately, matching up isky codes and all of that. So there's just so many different moving parts. Multiple people need to check these before they go out. So it's just a massive endeavor to try and put out something at scale, even just a couple hundred spots. I'm actually kind of reminded of a job that I had before coming to Misfits and Machines where it was a national brand with a bunch of different franchisees and they wanted to have a different promotion for every day of the week. We had to customize them for each franchise because they had different offers. The offer tag was in two different spots and again a different spot for a different day and a different location. So it ended up being, I want to say it was like 600 spots. We called it spotpocalypse. Yeah, we just sat in a room for an entire week just cutting everything, recording voiceover. It was a radio spot. And just the amount of organization to be able to do that and it was just mind numbing to have everybody sit there and spot check everything. So this kind of solves that where it's less of a jigsaw puzzle. And the mass customizer allows us to have more like creative input into what we're doing as opposed to just trying to prepare for this massive job to come through.
- [15:30]
B
What is possible with the mass customizer? What variables can actually be changed? Walk us through how it works.
- [15:36]
A
Yeah, absolutely. It's tied right into marketing Architects pipeline for post production. So we can swap out any voiceover line, any kind of graphic that we want, supers, different things like that, pull in a whole different shot for a specific spot or part of the spot if you'd like. And it's really easy. It's just a simple CSV upload. You basically just say the lines that you want to replace and put in the new lines and it'll read that CSV file and just kick out all the different variations. And it happens in almost real time. So it's much faster than that, you know, weeks long process. You can kick out hundreds of spots relatively quickly. And the nice thing about it is it's not just taking a line and replacing it just without any care. It's really surgical. The way it works. We're doing voice cloning just to maintain the actual way that the voiceover delivered the lines, not just the words that were said. And we're only replacing the spots where it's actually changed. So the entirety of the spot that's saying the same throughout all the customizations, it does stay exactly the same. And we do it with a lot of math, a ton of AI tools to just keep that original creative intent. And that way we don't need to go and re get everything approved and go through that whole process of checking everything just because it just works, it maintains all the same integrity of the original spot. So yeah, that weeks long process can just be a couple of hours.
- [16:54]
B
What else is on the roadmap for this product? What kind of personalization at scale do you think will eventually be possible in television?
- [17:00]
A
Yeah, so we're working on just tighter integration with marketing architects, post production pipeline, just tightening up a few things and then we're entering into pilot phases with clients where there's a natural fit. So that's the next step. See what we learn from those pilot tests. In terms of what's possible. The mass customizer really opens up a ton of possibilities, right, that we weren't able to do before. We may be able to do some scaling for addressable TV where we're checking household level targeting with a broader reach because the creative side can now kind of keep up with that. The mass customizer does one thing really well too. It's just swapping out those phone numbers and URLs, like the really boring kind of stuff. But that's maybe not as personalized as we would want something to be. So I think we can go and start getting really creative with weaving in kind of new levels of personalization that weren't really possible before that can uniquely tie in something about a person's demographic or their location, their life stage. And we can really personalize that in a way that's much more unique and interesting.
- [18:00]
B
Eventually you could almost. Could there be a world where a brand could look at what sort of personalization has worked for us in digital? And then at a household level you could start to apply some of that to tv. You think?
- [18:13]
A
Yeah, of course. I mean, that's the goal. You know, there may be some roadblocks in terms of how quickly we can deliver a spot once it's created to the network. But that's for them to figure out.
- [18:22]
B
So Matt, how are you imagining brands are going to use this tool? What sort of signals is your team going to be watching to measure success
- [18:31]
D
in terms of using it? The ideas are truly endless. Anytime a brand's message could resonate more strongly with a specific audience if even just like one element changed. That's a use case. Think different products, different regions, different storefronts. There's different offers that are customized across the country. There's honestly endless use cases. And that's just where the fun begins. Brainstorming all those different concepts of what it could look like. In terms of measurement. I think measurement always Comes down to two things. A, is it performing and B, is it having a brand impact? Are you hitting consumers in the right way? I think the worst case scenario here and something you wouldn't want to do is there have been cases, Lina, you said it earlier, where personalization goes wrong. That's definitely something we are very thoughtful in. And you can try to do some stuff in ctv, but how far you push it, like we wouldn't be calling out names of individuals on a television. You don't know who's actually on the other side of that screen.
- [19:29]
E
Hey there, Elena James.
- [19:31]
C
Yeah.
- [19:32]
D
Or even sometimes you can land nearby in your neighbor's house instead of the exact one. So there's just different things. But in terms of measurement, there's always the two sides of it. Right at the end of the day, is it driving revenue and how are consumers feeling about it? And measuring through brand studies, how is consumer feedback coming through through that personalization?
- [19:50]
B
So this kind of personalization on television is something that wouldn't have been possible on linear. It's I think an example of more opportunities that are going to be available to marketers due to ctv, which is kind of exciting, like the ability to use TV for mass reach, but also incorporate some of the capabilities we really like from digital. Ange, where do you think all this is going? What do you think is the future of personalization in TV and what other changes might be coming that marketers might be underestimating right now?
- [20:18]
E
CTV is just so exciting just in terms of the development of the space because it breaks the false choice between, between mass reach and customization.
- [20:29]
B
Right.
- [20:29]
E
You still get the attention, the impact of the living room screen, but now you have a digital style environment behind it that creates a lot of opportunity. What I think marketers are underestimating is how fast this is moving. CTV already accounts for nearly half of total TV usage in the US overtaking cable and broadcast combined for the first time. Ad supported streaming is growing as fast as viewers choose to tap into lower cost tiers. And that's a massive addressable audience in a high attention environment. The underestimated piece isn't just targeting. It's dynamic, creative at scale. What we're talking about being able to swap variables in a TV spot based on geography, time of day, audience signals, even weather. It was just not possible in the linear side. You could do it, but you put yourself out of business trying to do it. So tools like the mass customizer are going to make it possible to do it from one core asset and that's the unlock personalization that doesn't sacrifice the reach of mass media and done in a cost effective way. I think the brands that win will be the ones who figure out how to do that both at once. Maintain a consistent, distinctive brand identity while using CTV's data layer to make the creative feel contextually relevant. It's not an either or. It'll require some discipline to execute.
- [21:43]
B
Yeah, I think that's important to say, like, this isn't about just slapping on digital capabilities onto tv. It needs to be approached differently because of all the reasons why TV is different as a channel. All right, to wrap us up here and Matt, we'll start with you. What is the best example of personalization you've experienced as a consumer?
- [22:02]
D
Okay, I don't know if this is the best or just the most recent, but I'll let you all in on a little guilty pleasure. I love big chain food and Chili's is at the top of that list. They just started using the last thing I ordered in their text messages. So I am currently every single week getting, hey, don't you want some more fajitas? Come on in. And I like, I immediately start salivating and it's definitely making me want more chilies.
- [22:31]
E
I'll be honest, I love that about you.
- [22:36]
C
Been done with chili since I've known him.
- [22:38]
D
It's so good.
- [22:40]
B
And Rob?
- [22:41]
E
Yeah? I'll go next. Spotify wrapped, right? It's the most like shared, talked about piece of personalization marketing that I can remember recently. It makes me feel seen versus tracked. So I loved that about the Spotify wrapped functionality.
- [22:57]
C
Well, you're going to laugh, Alina, because mine, yeah, direct mail is an amazing thing and it's so old now we think of, oh, direct mail is just this archaic art form. But I still, to this day, whenever anyone asked me about personalization, it was opening the mailbox, pulling out a really nice piece of, you know, direct mail with a picture of my house on it and how much it cost. And I'm like, that is incredible. Do they have the FBI as their agency? Like, it's a little creepy, but it was powerful and I still remember it. And obviously it, it made an impression on you as well, except they screwed up for you, so that's not so good.
- [23:39]
D
But.
- [23:39]
B
No, but when they get it right, it's cool. My example is also direct mail, which is I got a direct mail piece sent to our office a couple of weeks ago and it was this new type of business card that's called a Taft card tap T. And they're very cool. You just have one of them and you can tap it on anybody's phone and it'll automatically put in your contact information, like Android iPhone. You can add your LinkedIn, you can add a photo so you're not having to just hand out physical business cards. And they sent me one that works. It's like fully functional, has my name on it. It's even in our brand color and has our logo. And I actually ended up having the marketing and sales teams order it. I'm like, oh, what a great way to share their product.
- [24:19]
A
I'm always drawn to. It's really simple, just that I live in Austin, Texas, and anytime there's like an ad on a podcast or on TV where it's just, hey there, Austin. It almost feels like they're talking to me. I don't know why, but one of the more unique ones, I actually got this last week. It was a card that I got in the mail from the company that delivers our dog food. We've got three large dogs and it was a happy birthday card to one of our dogs for a little discount. And I was like, oh, that's. That's nice. That's not like, creepy and invasive. It's just a nice little personalization. So that's one recently that kind of hit me as a good idea.
- [24:53]
B
I love that. I. Our dog turned two this last weekend and she has this app, I don't know if you've heard of it, it's called Fi. It's like Strava for dogs. It like, tracks her steps and there's leaderboards. You should get it for your dogs. Honestly, it's so fun. But they had a happy two years birthday message and they actually, it said powered by ChatGPT. They made her like a little song about her breed and, like, how active she is. I'm not gonna lie, it delighted me. I loved it. Anything to do with. Yeah, Personalization for a pet, I think is like, extra fun.
- [25:22]
A
Yeah. Pets are forgiven from all the weird targeting, I guess.
- [25:26]
B
Yes. Yeah, definitely. All right, that's going to wrap us up here. Thank you, Matt and Josh, for joining us today. That's it for this episode of the Marketing Architects. We'd like to thank Taylor De Los Reyes for producing the show. You can connect with us on LinkedIn and if you like the podcast, please leave us a review. Now go forth and build great marketing. Marketing architects.