Sydney Sweeney & American Eagle: The Double-Edged Sword of Celebrity Branding

[00:00:00] (teaser) ​

[00:00:00] Todd: how could you be spending 10, 5, 10, 15, or $20 million a year in media, but not have a sophisticated mathematical model to help adjust?

[00:00:12] Kelly: you give me more money, I have mathematically backed and statistically strength relevance that shows you, you will get more than what you give me back in returns, and that becomes really powerful.

[00:00:48] Kelly: (interview) Everyone. Kelly McGuire with Bougie Day Digital. Excited to get into today's topic.

[00:00:52] Todd: Hey everybody, it's Todd Juno, Kelly's partner. It's Monday. We're gonna talk about M and mss, but not the candy. Well, maybe it is candy for marketers. You know what, if I was a marketer, this would be my candy. This would be my chocolate crack that I would have to have, which is the 3M. The MM. . World that we are working in, in a variety of ways we've talked about before, but we've got some new evidence of some things bubbling up that I think, can refresh the topic we could talk about today.

[00:01:20] Todd: , And it's always good to just talk about, you know, what is in this case, media mix modeling, not marketing mix modeling, but in our particular field of work media mix modeling. So, Kelly, today I am gonna pick your brain about that.

[00:01:33] Kelly: Yeah. Before we go into that, one point of clarification as we go through this. So media mix modeling, marketing mix modeling. MMM, when I look at it, it's also, it's kind of a catchall term for essentially correlation analysis, right? Being able to look at the relationships, the trends, and the predictiveness of.

[00:01:50] Kelly: One thing leading to another, and the strength of confidence that they are tied to each other, or, causation and correlation all of that. So because it can get confusing, and this happens in other areas where there's mixed media modeling is a description of a way to do that, but it's really just the correlation of the relationships between metrics, outcomes, efforts, all of that.

[00:02:09] Todd: Well, thank you for defining for the audience that that is how we look at it. Probably how it should be looked at. 'cause there is a lot of slang, language and acronyms in our field of work. , So we'll table set with that, that MMM is sort of this broad sort of label, but it really sort of all ladders down to a very specific practice of interpretation of things.

[00:02:31] Todd: we'll dig into that. , I've got five or six things to ask you about. , And so we'll start with. Myth busting. , There are maybe some myths about MMM, well, I've read some myths about MMMI should say, whether it's accurate or true or false, or capable or incapable. , What myths have you encountered, , that you think are true or false, or potentially would prevent a brand from investing?

[00:03:00] Todd: In a broader strokes, MMM activity.

[00:03:02] Kelly: I don't know if it's a pervasive myth per se, but I can say some of the things that make it valuable or useful or the way that I tend to try and use it. I mean, first of all, like anything else, it relies heavily on the accuracy and the consistency of data. The whole point of being able to leverage correlation analysis is the recognition of patterns and the reliability of those patterns, repeating and staying consistent and regular.

[00:03:28] Kelly: And so being able to look back over time and account for a lot of variables is really critical in whether or not it works or not. It this is why the whole idea of correlation becomes really nice because direct attribution of media is very narrow in its ability to. Measure outcomes, and we're getting less and less visibility into that as we lose trackability and, , data privacy and all these things comes in.

[00:03:54] Kelly: And so knowing that media is one piece of a big puzzle, being able to track those patterns over time. Really helps understand the role that it plays in the likely outcomes in a larger sea of things that you can't always account for. But again, that requires consistency in the setup and the data. So a myth, if you will, that I would look at is that really anyone can just do and apply an mm m right away and it's just a matter of paying for it or, getting a data scientist or a company to come do it.

[00:04:24] Kelly: You can have the greatest tool in the world and the greatest data scientists there are. If you don't have the fuel or the data that will arm you with this, the best you can do is start to build towards a future outcome with a certain level of reliability. And I would argue there's still value in doing that.

[00:04:40] Kelly: The sooner you go towards that, the sooner you can get to a reliable outcome. But it's not just useful from day one unless you have strong historical data to arm and train. It's kind of like. The LLMs right now in sort of the AI world, right? It's all about the training data and the information that informs it, that it can train on to give you the right outputs.

[00:05:00] Kelly: MMM works the same way. You need consistent structured data, and the better you can do, the better you can collect it, the better off you're gonna be, but also you can build towards getting that even cleaner and stronger. And so the relationship between all of your marketing activities. I guess maybe that's another one that I can look at as a myth is MMM doesn't only look at media outcomes.

[00:05:21] Kelly: That would be the media mix modeling, but the marketing mix modeling can account for other variables, essentially, really anything that contributes to a certain percentage of your outcomes, whether that's. PR impressions or social media activities or seasonality there's all these components that can be plugged in if you have access to that information.

[00:05:40] Kelly: It's consistent, it's predictable, and you can plug it in in a realistic way. And so that is the correlation of all of the thing moving parts that are creating results in the ability to start to measure and recognize patterns that indicate likelihood of what can or is happening in relation to those.

[00:05:57] Todd: Okay, so you talk about a lot of the parts and pieces that are needed. . Paint the picture of a brand journey for me to answer the question of when does it become worth it to invest in the effort that it takes, and is there a point where you're too small before that starts? So, think of it this way.

[00:06:15] Todd: You started a brand, , in your case a stocker equipment accessory manufacturing brand. So you make balls and jerseys and gloves and. Shin guards and all this kind of cool stuff, and you just launched and you've got this state-of-the-art technology that's making the best products 'cause of the best materials ever.

[00:06:32] Todd: , But you just started the company and you don't have enough data to train it, but you've got a lot of money. Talk about when, where, and how the ability to actually use it works. You talk about training, so probably can't do it as a. Startup company, right? 'cause you have to have a lot of data.

[00:06:51] Todd: You might not be able to do it 'cause you don't have enough money. Like what are the moments when a brand should be thinking about, okay, now I have what it takes, data, money, time, resources, whatever, in order to implement something like this so that I will get consistent, accurate, trend worthy information that I can make decisions off of.

[00:07:11] Todd: When does that moment happen?

[00:07:12] Kelly: I think it's a unique answer to right now versus what my answer to this would've been maybe even like a year ago. And a lot of that is done with, AI tools right now and what it's allowing for. I think the question is better broken up into. The scale of correlation modeling that I would access because. With AI and the tools that are in there right now, I would argue that you can start to look at it in this frame of reference and frame of pretty much right now without major costs or investments. If you have the right mindset or skillset or the right people working on your business, that's kind of a different thing to, go into, but the ability to kind of look at the patterns and and relationships between moving parts.

[00:07:57] Kelly: It can be done pretty well within the systems that happen right now, but that's a very loose, informal way of doing it. That's just a matter of cleaning up your own data and your own media things right now. That's something that, that I do for our clients a lot right now. There is a next phase of that, which would've been the, full answer I would've given say a year ago before some of this, which is a true.

[00:08:17] Kelly: Invested tool-based data, science driven, real full fidelity mixed media model that costs with that. The answer to that would be when the outcomes and the information at that full investment are, I would say maybe five to 10% of what my total spend is. I wouldn't, I wouldn't spend six figures plus on a really baked in mixed media model until.

[00:08:43] Kelly: I was spending at least 10 times that much so that whatever I was learning and informing and predicting and growing was really going to be leveraged and yield kind of the outcomes from that. And so if I'm spending a million dollars or more, there may be a reason to start to kind of look at, okay, I wanna make sure I'm spending that and if I have ambitions to grow that million dollars to two or three or four, and to know kind of what those maximized returns are, then it starts to create a good balance of, of what.

[00:09:10] Kelly: It can guide me to do because again, there's so many different things that you can spend the next dollar in marketing on. As much as I hate to say it, like the answer isn't always another dollar into Facebook or another dollar into search. It might be more money into better creative a, or a promo, or a discount or a PR thing.

[00:09:30] Kelly: There's all kinds of things that work together to create outcomes and when you're just measuring things like platform driven. CPAs or click through rates or those things like that, you can get so focused on the efficiency of a single channel. You can kind of lose the forest for the trees and understanding how maybe search, is incredible, right?

[00:09:50] Kelly: Search has always benefited from being like the last touch, the last thing that people do. And so it's always looked really good. We've always made the joke about, well, how do you know someone would've either done that search or clicked on your result if they didn't see your CTV ad leading into it? It recognized your brand before.

[00:10:06] Kelly: That's a really hard thing to know and measure, and it's historically been really hard. Well, mixed media modeling really done well can start to show the balance of things that are upper and middle funnel and how they lead into and grow into the bottom funnel that have always taken credit and so. When I get to a stage where I'm activating multiple channels and I have multiple things that influence my user journey and I need to understand how they all work in conjunction to each other, that's when I would get really, really serious about making those types of investments, , to have that visibility.

[00:10:37] Kelly: , Anyone that I would be paying to work for me. And work on my media. I would at least like to be hearing that the thought process around correlation and bigger picture and how things work together.

[00:10:49] Todd: How much media, how much media do I need to be spending? So you just said it's gotta be a factor of 10, right? Well, we know there's tools out there that we work with, and their cost could be a hundred thousand dollars a year for. License to use their tool. Right? So does that mean that somebody's gotta be spending a minimum of a million dollars in paid media?

[00:11:08] Todd: Be specific if

[00:11:09] Kelly: I think, like I said. If I'm spending enough money to where I am paying somebody else to be running multiple channels of media for me, then I think at that stage, whoever I've. Entru with that responsibility should at least have the thinking and the ability to use generalized tools in that way to kind of not silo any channel out, but to be able to have a viewpoint on how things work together.

[00:11:33] Kelly: How, awareness leads to traffic leads to. Conversion and kind of the relationships and when promos have an impact and pricing has an impact, and when seasonality has an impact, the ability to look across multiple channels. If someone else is running that for me, regardless of my spend, I'm gonna assume my spend is high enough to justify paying somebody else to do that.

[00:11:53] Kelly: So let's say tens of thousands, hundreds of thousands, whatever that is. Certainly when I'm gonna make the investments into the technologies and the real sort of advanced stuff, , I'm probably gonna be into the millions of dollars of spend. Realistically, mms, I believe were, traditionally applied to.

[00:12:09] Kelly: 10 million plus clients that really have huge data sets and things like that. Like I said, those thresholds are getting smaller and lower. As things get more commoditized, data becomes more ubiquitous. , Just the access to the information, the ease of using it becomes less pure like. Figure, high $500,000 a year data scientists, or you can't do it type stuff.

[00:12:32] Kelly: you're getting into the tools, you're getting into esia models and regression modeling and these really advanced mathematics that are available, I would do that when I'm spending millions of dollars. . Certainly if I have ambitions to grow and scale and I have a business that is driven by a lot of variables that aren't just put a dollar into Google and get a dollar out, like when I start to have to think about pricing and distribution and promo calendars and partnerships and influencer content and PR and all these things, I would need a visibility on.

[00:13:01] Kelly: What the right mix and the outcomes are for that, and I absolutely would put it into place if I had a business where my marketing team had to make this case to leadership and A CFO or a board who pretends to not really buy into the marketing impacts of things, the ability to kind of really show that data and that correlation and that forecasting.

[00:13:23] Kelly: Is a big overcoming of that barrier of convincing people. If you give me more money, I have mathematically backed and statistically strength relevance that shows you, you will get more than what you give me back in returns, and that becomes really powerful.

[00:13:37] Todd: So I'm gonna pick on what you just said a little bit in the industry. I would say we're a small agency, so we don't come across the big brands, but let's say the small to mid-market size brands that are spending one to 20 million annually in media, I would argue based on what I've seen, less 25% or less of those brands with budgets of that size are deploying anything like this.

[00:14:02] Todd: Because when we encounter them, most of them don't have this. So why do you think that is? Well, how could you be spending 10, 5, 10, 15, or $20 million a year in media, but not have a sophisticated mathematical model to help adjust? Decision logic to where things contribute. You said correlation, contribution, those kinds of things.

[00:14:27] Todd: Like first of all, a, do you agree that it's a much less percentage of people using it than are using it. Would you agree to that?

[00:14:35] Kelly: I would absolutely agree to that, and I would take it one step further. And I would say the smaller majority of the ones that are probably are not doing it either, right, or as effectively as they should be because there's a lot of problems on that front that I could get into as well.

[00:14:47] Todd: we know that 'cause we've taken over some of those. But that being said, if that's the case, if you're a marketer and you knew that the data science is good enough now. and the cost of the tools and the resources to implement are reasonable. If you're spending 10 or $20 million a year in media, why would you not have this in place?

[00:15:07] Kelly: it's hard and you have to pay for it. It takes time.

[00:15:10] Todd: Let's assume you can budget it, right? If you're spending $10 million a year on media buying. you're a pretty good sized marketing budget, right? cause that 10 million is only a percentage of the overall marketing budget. So if the overall marketing budget is $20 million, that's 10% of revenue, then you're a pretty good sized company.

[00:15:28] Todd: Why would you not be doing that?

[00:15:30] Kelly: I mean, I absolutely think you should. I think the reason you wouldn't be is because what I said is it costs money. But at that point, I mean in the front. Actions of it, but it does still cost money. You have to commit to investing in it. it takes time and it takes expertise. So you have to have a champion or a partner who cares enough to bring the, solution to the table because it also creates more accountability, , which not everyone wants.

[00:15:53] Kelly: But I think obviously that should exist more, , because for better or for worse, more accountability is more scrutiny, is more responsibility, is more outcome. there's a side of, this industry that doesn't always like to do that. , You know, I don't wanna go too far down that slippery slope, but, you have to have someone who is looking out for the best interest and wants that and can action it and can use it because yes, it costs time and it costs money and it takes expertise and it also requires.

[00:16:19] Kelly: Commitment to seeing it through, you are going to have to redo the way campaigns are set up and structured and data's collected, and it just requires so much more cleanliness and responsibility and fidelity of work. That again, I would argue if I was in charge of round, that would absolutely be what I was mandating.

[00:16:38] Kelly: But there's so many moving parts and there's so many responsibilities and people tend to wear so many hats these days. It's just really easy that that doesn't always come across as the next biggest priority or, the next right thing or, and all of that. So, , I mean, I think there's a lot of white space for these tools or these capabilities to be being sold into these brands, but every one of these brands is being sold 1,000,005 different things than any given time.

[00:17:02] Kelly: It's just a matter of the breakthrough and the commitment and someone really, going for it. But I think. I would distill that down into your easy chat. GBT is, it's not that easy to do It takes commitment and work and so therefore a lot of people don't get around to it.

[00:17:15] Todd: I'm just fascinated by why somebody wouldn't want to know the level of transparency, clarity, accuracy, for the most part, accuracy, dependability, , of what these models can produce. Again, we know because we've seen what they're capable of. So I have, I have a two part question for you, or maybe they're just two separate questions.

[00:17:34] Todd: , if you're a brand. Can you be specific on what do you actually need? what does the brand need to say? I checked the box, I have these things, these items, whatever it is to implement some sort of media or marketing mix.

[00:17:48] Todd: Modeling process, technology license. be specific and like, here's what you actually need to do this.

[00:17:57] Kelly: as simply as I can put it, is you need consistent data streams that are segmented in increments. So for example, I need to know my. Facebook ad spend on a daily, monthly, or weekly basis, consistently going back months or years. and it has to be consistent because again, you have to correlate it out.

[00:18:19] Kelly: So I need to know my sales or my revenue, or my ad spend, or my impression level, or my click at a granular, consistent level. So I need it by channel, by tactic, by timeframe. Going back as long as you can possibly go, because the longer you go the more it can account for things like seasonality or other disruptions in the business because it flattens it out over time.

[00:18:42] Kelly: 'cause it looks at the relationship over time and it looks for patterns and there's a lot of anomalies that happen over time. but you, you need the consistency of the data in whatever slices you have. Again, it could be my channel, my tactic by timeframe. But it has to be the same data in the same format for as long as possible.

[00:19:00] Kelly: And ideally, you would have that data available for all of the things that have a meaningful impact on the business because let's say for example, simplified out, let's say you ran search, social and programmatic at an equal level, and you spent an average of a hundred thousand dollars a month per channel, per tactic, going back over time, if you only had the data for two out of those three channels.

[00:19:24] Kelly: You're gonna have a huge blind spot on something that could have been moving. The reality around that, the model can't account for, and so it might be giving you inaccurate or unreliable outcomes because it has too big of a blind spot to account for it. Now again, the fact that it's blind and consistent doesn't render it useless.

[00:19:43] Kelly: It limits its overall value by having that blind spot. So you need to be able to have that data available for the most important, impactful things that likely contribute to those outcomes. Because without that, you can come to the wrong conclusions in the models, and that's where it becomes difficult, because if a brand has been working, Three different agencies over the course of the last three years, right? Every one of those agencies that they were working with likely ran different channels, Maybe they didn't even own the accounts, so they don't have the data. They named them differently. They used different tactics. And so now you don't even own that historical data because you've moved around partners and no one tends to remember to collect and store all that in a really useful way.

[00:20:26] Kelly: And so again, your worst case scenario is if you don't have that. You can start from scratch, from day one, but now you have to work with all your partners with a clean vision, a clean structure, so that everything moving forward is consistent and you're building your own history to increase the confidence over time as it it does that pattern recognition.

[00:20:43] Todd: All right, so here's the tougher part B of that question. Let's say you have all of that, ' cause you've been spending a good sum of money on paid media.

[00:20:53] Todd: you're the marketing director for a brand that's spending $5 million annually in paid media.

[00:20:59] Todd: You don't have a media mix modeling tool or license or the, the skilled labor to process and manage and run something like that. How are you making decisions on where to invest media dollars in the absence of that?

[00:21:13] Kelly: it just tends to be channel by channel in individual. It tends to be based on whatever your meta metrics are or your keyword research and your share of voice and your, CPAs or your cost per clicks or whatever. Platform specific information is being given to you. It tends to be a very siloed approach in that case where it's, I'm maximizing this channel and I'm getting this channel and I'm running this channel, versus being able to look at the bigger picture and how they all sort of ebb and flow together.

[00:21:40] Kelly: that's what I would think would be the norm. If you don't have that broader, kind of higher level of visibility on it. I'm careful in making that statement ubiquitously because. Mixed meeting model is not the only way to not silo your, media execution. But I would say it, it doesn't tell as much of a story on a lot of those variabilities.

[00:21:57] Todd: Let's close with an example. , I know that you have had, or we have had the ability of working with a couple of different brands, , let's say in the last 12 months where a licensed MMM technology product has been, , applied. , And you recently got to see. Some interesting outputs of that.

[00:22:18] Todd: , Can you talk about, obviously we can't give away sensitive client names or any information, but can you talk about sort of in general, what did you see as the output? How did it impact, you know, the larger client side conversation relative to, you know, how their planning for their next year, , how that impacts us as an agency for them?

[00:22:39] Todd: Like can you just talk in general sense of. When it's applied and it produces, visible outcomes, what does that look like in a good case scenario that you've seen recently? And, how do you see it sort of now that that's happened? Like does it get just put to the wayside for a while and then you start it again?

[00:22:58] Todd: Or do you keep it running? Like what's a good case scenario look like?

[00:23:02] Kelly: Yeah, so to to protect client outcomes, mix and match, , benefits and, outcomes that I've seen across different campaigns over time. , I've seen it do a lot of things. I've seen it create greater confidence In the ROI of media, and investments which have led to increases in the willingness to explore and budget by way of showing that it tends to be modeled on a bell curve.

[00:23:25] Kelly: And a lot of times what we show is that we're on the, left hand side of a bell curve, which means we're not even at max profitability. So you're not even extracting your absolute best return. Then it often shows that even on the right hand side of the bell curve, you may not be at your max profitability, but you're still very, very profitable for a good way down the right hand side of the curve.

[00:23:44] Kelly: And it tends to tell you that you can, spend a lot more than you do and still extract positive ROIs on that.

[00:23:50] Todd: When you've seen that, have you then directed a brand to actually spend more money as a result of that? Where that output and that bell curve actually ha, no pun intended, rang true.

[00:24:03] Kelly: Yeah, absolutely. And, and what it does is it creates that level of confidence at the CFO level that it tends to be where you're making a prediction of returns and then you actually have a measuring barometer of whether or not it's happening that way. And so you can say if you. Invest X, we expect Y.

[00:24:19] Kelly: And it's not a blind like you do it, you run it and you measure it after the fact. You can watch if you're seeing that and the metrics and if it's following those predictions, and you can backfill it from hypothesis to actuals and start to see how far off or how close to those you are, and you can refine over time.

[00:24:34] Kelly: This is the beauty of these systems is they're learning systems that constantly improve. They give you projections, you run things, you input new data, it tells you how close you are, and it learns and gets better and constantly improves.

[00:24:46] Kelly: So that's where it tends to be the best around investments and forecasting and predictions.

[00:24:51] Kelly: But I've seen it start to smooth out seasonality. , There was one instance that we just had that suggested that. Media was being run, very start and stop, and you were missing the beginning in the end of the year. Whereas if we stretched it out longer, you could carry a halo effect from the end of the year into the beginning.

[00:25:07] Kelly: Instead of starting from the ground floor, you would create a little bit of draftable momentum that would increase your ability and your max time periods. it shows between over and underinvestment in certain channels where. Your lower funnel tactics tend to get the greatest credit, but your higher middle and upper funnel tend to drive and allow more opportunity so you can get more retargeting.

[00:25:29] Kelly: So shifting budgets from lower to middle and upper funnel and looking a little bit further out on those ROI metrics, it tends to, guide those types of things. it can do audience modeling. , We've done it for offline measurement where. We don't have a direct digital attribution signal.

[00:25:45] Kelly: And so how do you create connectivity between media investments and someone, you know, walking into a store or, buying something offline and connecting the dots between those and creating a greater sense of driving those types of things. So. I'm trying to kind of think about all of those outcomes, but it, is so illuminating in what it can tell you.

[00:26:03] Kelly: It can, it can tell you promo sensitivity when you should be able to run more pro promotions. , Versus when it's not price sensitive. It can really just give you this understanding and this, this variability of when you're under and over investing, when there's different time outputs, whether there's different audience elements.

[00:26:20] Kelly: it gives a great guidance. Again, like everything else, it's only as good as the inputs that's going into it. It's an evolving, growing system. So again, you have to have a champion and a team of people that are both willing to continue to leverage and use it and inform it. Also react to it and adjust to it within reason.

[00:26:38] Kelly: You don't wanna follow it blindly. You want to use it as a input of decisions of where should my next dollar go? And you want to measure, the mixed media model tells me that I am under invested in these channels. My platform metrics tell me I have this much opportunity. I should spend a little bit more in search, or I should put a little bit more in meta, or I should capture these keywords.

[00:26:59] Kelly: it's just a really good higher level. Input that allows that decision making to, to be really strong. But more often than not, it's just telling you what's my point of diminishing return? How much more can I go, when do I need to pivot from lower funnel to nurturing more of the upper funnel?

[00:27:16] Kelly: gives a lot of insight around those types of things.

[00:27:18] Todd: Alright, well, If I'm a marketer and I'm listening to you and I'm not doing this, I might start to just bash my head against the wall because I can't imagine being responsible. Well, I'll go back to just as a quick closing note, my life on the brand side, when I was in charge of a large marketing budget and I did not have this, I didn't know.

[00:27:41] Todd: Couldn't tell, , with few things with exception to trackable phone number or maybe a unique URL or Vanity Code dating myself, , that anything was working and was there any correlation between any of those things you really didn't know? , And so if I was doing that today, and I knew that these tools were available and the technology was available, and the mathematics were available.

[00:28:04] Todd: I can't see how I would not be doing this,

[00:28:08] Todd: So we will wrap on that. , , be safe and I'll talk to you next time about some other interesting digital marketing topic.

[00:28:16] Kelly: Thanks everyone.