[Music] [E.J. Freni] Well, thank you all for coming. My name is EJ Freni, I'm the Chief Revenue Officer of Claravine. We are a data standards company, which I promise you'll know what that means after we're done. I have the best job because I get to work with some of the biggest brands, their agency and SI partners in the world. And through that, I've got to meet our good friend, Mark, who's with Colgate. [Mark Zomick] Hi. Whoa. I forgot to tell him in the back, I have a pretty loud voice. Hi, everybody, I'm Mark Zomick, I work at Colgate. My remit is around media activation reporting for Analytics, and through that role we've managed to discover Claravine, and you'll hear of some of that journey today. So we could have named this presentation a lot of things. How many folks, show of hands, were at the Keynote this morning and heard Hilary Cook from Marriott speak? So you heard her say taxonomy is sexy. That should have been the name of this presentation, which, she had me at hello the minute she said that. Or we could have named it Bringing Control to Kaos, Kaos with a K, Mark. Right. It's the Get Smart reference. Sorry for those of you-- Yes. I'm going to date myself as well. Show of hands, how many folks early on remember the game Oregon Trail on the Apple IIe, okay? So for those of you that were not an Oregon Trail user, it was a simple game. You had to get from one side of the East Coast of America to Oregon. Perfectly linear, right? Well, inevitably, your oxen would die crossing a river, your party would get dysentery, you'd eat bad blueberries or something. I had a very close friend of mine that's on the marketing side tell me this is how it is to launch a marketing campaign. So just to kick things off, I'm going to talk a little bit about what we see when we talk to our brands and our partners in the world. And then Mark's actually going to show you how Colgate has evolved their data standards practice globally with some amazing success. But many of you marketers in the room, you've got a big challenge ahead of you, right? Or you have a big challenge in general. Just internally, launching a campaign can be very complicated, arduous. You've got silos with your teams. The average marketer has 90 to 120 SaaS applications. So even if you have integrated Adobe almost full scale, you still probably have dozens of SaaS applications across your marketing technology org. And if you're a global marketer, even if you're a regional marketer with multiple BUs, you've got to deal with this data chaos at the global level, the regional level, the local level with different needs from different teams across different business units using different technologies. And so this misalignment can lead to massive problems for everybody. Back to the Oregon Trail, before you crossed the river with the oxen, maybe you didn't realize that the wrong creative got trafficked. Or your campaign goes live and you don't even realize that things were named or the tracking wasn't accurate for six weeks into your campaign, right? So at the best case, you're probably losing some money on that data loss. Worst case, you've got a lawsuit on your hands because a piece of creative went out that wasn't supposed to go out with an offer in a region that should have been in that region.
So that's the internal problems, and process issues, and silos that most of you in this room are dealing with, but then, of course, we're all dealing with all of these market dynamics. And I'm not going to go through all of these at the top. We've been dealing with this for years, the continued signal loss, the massive evolution and transformation we all had to go through in the pandemic. Three years ago at Summit, that's all everybody talked about was this massive digital transformation that people had undertaken in months.
The proliferation of channels, TikTok wasn't even really a channel that people were investing in five, six years ago, now it is. The walled gardens of retail media. Now more than ever, you have to make sure that all your data, when you launch a campaign or you're appending data to a piece of creative, you have to know exactly what you did before it goes out into these environments because you're only going to get a certain amount of data back. But we have heard nothing but AI these last two days, right? It is one of the most exciting things, if not the most exciting thing of our lifetime, but it's also going to come with challenges, right? Everybody in this room, I guarantee was using some form of AI five years ago, four years ago, but then the ChatGPT moment hit in the last two years. All we have been talking about are LLMs, agents, GenAI. And again, I could not be more excited for this, but it's also going to create a treasure trove of data that we can all take advantage of. But if we don't have a system in place and we don't have a process in place to capture that data, collect that data in a standardized format, it's going to be the old garbage in garbage out adage virtually for any system that you use that is going to be leveraging GenAI.
So that's the problem. Well, what are data standards? What are we here to talk about? So data standards, the most simple thing, Hilary stole our thunder this morning. Taxonomy is sexy, right? One of the most important fundamental things that you all can do when you leave this room and you leave Summit is just think about your data standard strategy and how important that data foundation is to everything you do. Optimization, measurement, analytics, business decisions. But when we use the term data standards, it can be many things. It could be your common data language, your data dictionary, your taxonomy. One of the most important things that we always talk about with data standards is metadata. That's another term, two years ago, you hardly heard it at Adobe Summit or any conference. I think I've heard it 1,000 times, right? Metadata is the context around your data. It's the data about the data. The best example that I've heard is you might have a giant sheet of seven digit numbers, but you have no idea. Is that some random ID or is it a phone number unless you have metadata associated with those numbers? So metadata is going to become more important than ever, especially in this world of GenAI. The other reason that data standards are so important, more and more marketers, I'm sure many of you, if not all of you in the room, were using campaign IDs or creative IDs. If you have your data standards in place, you can start to append all of this data, your data dictionary, your taxonomy, to those simple IDs and have a treasure trove of data to look into to make business decisions. So before we actually show this in practice, just a quick example, everybody here, unless you're local, you probably flew in, and the airlines are a great example of an ID that is actually used to extrapolate data standards. There's a PNR, which is your passenger name record. And pro-tip for Mark, if you ever have trouble at the airport, never mention your SkyMiles number or your flight number, mention your PNR, and they'll think you're a pro right away. But with your PNR, from that one number, any airline can figure out who you are, dietary restrictions, the seat you're in, the pilot, your last five destinations, all of that through one ID. And all of that because in the back end, they've got a great data standard system, and they're using this to do personalization at scale every single day, some better than others obviously.
So before I hand it over to Mark, data requires collaboration and connection. So just as you could have an amazing process and you could have an amazing taxonomy and standards in place, you've got to think about collaboration. Because one of the things that we run into all of the time, you could have the best system and process in place, but just as much of a problem as those 90 to 120 SaaS applications that are in your tool chest, also you've got other partners. You've got partners that might vary by region. You've got systems integrators, consulting firms, your media agencies, which we're going to talk about a great example of alignment with the media agency here. You need collaboration. And so our whole point here is that you really need a system in place to create that collaboration and get the needs and the collaboration and the data standards propagated out through all of these teams in a uniform way. Otherwise, you're going to have some real challenges. So Mark is going to actually walk you through the evolution that Colgate's been on, and then we're going to wrap it up in a little bit. Thank you, EJ. Thank you all for coming.
Taxonomy is a hack, right? Essentially, putting a lot of metadata in a title, in a naming box where the systems aren't governing it, is essentially a hack because all of these systems that we use, the one thing they probably have in common is a free form text field. They don't care about it, so they let us do what we want. What do we do? And I wanted to find the term of, as we've been talking about, metadata and taxonomy is a pre-agreed upon naming convention, or we use those words interchangeably naming convention that allows us downstream after activation, after one system touches the next, which touches the next, which we're going to talk about in a second, but it allows us at the end to know, that's where that came from, okay? We found ourselves, and this is not a Colgate specific thing. This is really pretty universal, consistently fixing metadata when it came back to us after activation. It was manual. It was time consuming. My friends at VML may know that the lead media director on the account would spend her weekends doing exception reports just to get that data to align afterwards. It's not where we want to spend our time. And because we really couldn't get over that hump, our analytics was limited. So we recognized we needed to own this change and make this change, and I really try to balance this revolutionary versus evolutionary because this was really a key to what we were able to do. The change we made in the process was almost nothing, but the change we got to our data was and is and continues to be really unlimiting in terms of what we can report, right? And that was one of our success factors, which we'll talk about, was that we didn't-- I've been through all host of I spent most of my career at media agencies, of all host of media agency transformations. And transformations are hard and change is hard. But at the end of the day, the transformations that always work are the ones that change the least amount of what people are doing in their day-to-day.
So there's fragmented reporting.
Each region was doing something differently. But at the end of the day, we're really trying to balance the revolutionary result with the evolutionary change to the impact, okay? Another term that I'm going to define here, and it's a little controversial. How many people work for agencies or have media experience other than my friends over here at WPP and VML? Okay, here's the secret for the rest of you. Okay? It's customary and pretty much all media operations around the world that you get to grade your own test, right? And the teacher doesn't come in until after you've rewritten the essay based on what the essay was supposed to be. Meaning that the buyer gets the data from the buying system, looks at all of the data coming out of the report, and corrects the paperwork mistakes that they made on the inbound. That's really universal, right? So the buyers are going in to say, "Yeah, I remember. I did that wrong. Now I have to fix it." So when the buyer tells me your reporting system doesn't match what my reporting system says, "I guarantee you there's an Excel spreadsheet somewhere with a lookup and a replace to match everything up," right? So we define, and my remit is, globally consolidated automated reporting, which is defined from the time the buyer's hands leave the keyboard until the time I can see it in our dashboards, there is no need for human intervention.
Every agency I've gone into for the few I guess I've worked for the big three or however many there are. Sure we're doing automated reporting. Can you show me? They log in. They download something from a UI. They bring a CSV into an Excel, and they run some macros. That's not automated reporting, okay? I'm totally not exaggerating. At this point, in all of the times that you've ever been to presentations like this, and I've been to many, there is a moment where everybody takes out their phones to take a picture of the slide. I wanted to be in the shot when you all did that...
So that now when you take this picture, maybe you'll remember me for a second. And this is the mess, okay? Or this is the mess. And honestly, this slide may be 15 years old with very little change, okay? This is a media operation. And a really good media operation, really good. And we love our agency partners. As a former WPP employee for 10 years, I have a very soft spot in my heart for our partners at WPP. A really tight media operation is going to input a media plan between idea and activation five times manually, hands on keyboard in a different format. I'm going to have my idea, then I'm going to put together my tactical plan in a flowchart, then I have to probably get an authorization to buy, which is different than the flowchart, then I actually have to go into a buy management system. And if in the US, it's Prisma or something like that, but globally there are many, many more systems. And then if you don't have the US version of the Prisma to CM connection, you're now putting it into a traffic system. Five times manually from idea to execution. That's without the back end change that I just talked about that these buyers are making, right? I want to say, this is an accident waiting to happen. This is 1,000 accidents waiting to happen every day. We spend a billion dollars in a year in advertising globally at Colgate and/or reported over a billion dollars a year in advertising at Colgate. And at the end of the day, we're spending that money a nickel at a time, right? You have to understand the throughput of that so you can able to report out there. So again, from the beginning, I have an idea. What does that plan look like? That's really a high level. I'm going to come up with an audience like, I don't know, whitening moms or something like that. And you create personas and a brand plan. But then ultimately you have to get to a tactical flowchart that has money attached to it so I know what my daypart mix is. Then again, you're going to have to do an authorization, usually by quarter, by vendor, depending on the size of what you're doing. Then all of these individual systems are separate standalone systems. The TV systems and the digital systems don't speak to each other, right? They're done completely differently. You'll see legacy systems. That's just the way it works.
And each country has their own way of activating. Sure, we can all use Campaign Manager collectively globally, but then you're going to have, you also have Meta, you also have other ad servers that are going on. And ultimately, at the point of execution, a number of things come together, okay? If it's done correctly, at the point of execution I'm going to have my, let's call it my campaign, right? The overarching campaign idea. I'm going to have the tactical placement, which is where the negotiation with a particular vendor happened. Then I'm going to have an audience, right? How did my whitening moms translate into my deliverable media audience? And I'm going to have creative. I have four things that are going to converge at the point where each one of us is served up an ad on the screen. And if you have all of those things organized in such a way, in a homogeneous way, in a globally homogeneous way, in an organized way, you're going to have a tremendous amount of data that's actionable. How did my creative work against my audience? How did my audience work against my creative? Which is different when you flip it around. And all of those things happen at the point of execution. Ultimately, the analysts need a seat at the table when you're having these discussions about taxonomy. Otherwise, when it comes back through the pipes, you'll never be able to reconcile what the analyst needs to be able to analyze, right? The analysts need a seat at the taxonomy table because without that, they're really going to not have the value. And that goes for onliners. That goes for Marketing Mix Modeling. That goes for whatever you're going to do. The more the analysts can drive what's in the taxonomy, the more you can thoughtfully approach your taxonomy. I don't really need my creative execution in my placement name because that's going to converge at the point of execution.
Ultimately, the idea is to create the connected view. So if you remember from the previous slide, this is a generic purchase funnel. You might have a McKinsey curve. Every brand has their own flavor of what that purchase cycle looks like. A McKinsey loop. Everybody has their brand, right? And ultimately, we know a lot of information once the user clicks on an ad, once the user served up an ad. But the question is, how do I take the information on the click and tie it back to my whitening moms and my highfalutin media plans to be able to establish the tactical media audience that I delivered is really this ephemeral media audience that we thought about in advance. So the first thing now is in our journey, we establish data and taxonomy ownership. This seems so simple, but this is where a lot of the challenges come, and I think that our approach...
Disarmed the challenges. So I'll give you an easy example.
I'll do it anyway.
Your finance teams decide which countries belong in which regions, right? And as much as there might be people in Washington who think Canada is the 51st state and is part of the United States. - That's right. - No judgment.
As much as that might be true, the finance team says Canada is Canada and it's part of North America. And nobody, no media planner is going to now say, "No, I think Canada should be the 51st state." It doesn't happen, right? But that's true on a lot of things that we don't consider, right? Products. I got a request a few weeks ago for our Latin America team. They wanted to add a product to our global product list that we have stored in Claravine. What product did they want to add? It was called ETB. So in order to add a product, I'm the gatekeeper now because the only way to do it is to have a bottleneck, is at least for now, is because otherwise you get a lot of things and here's a perfect example. I go, "I want to see a picture of the product." In order to get a product added to the list, I want to see a picture of the product to make sure the product really isn't in the 14,000 products that Colgate-Palmolive produces. It's not really there already. So she shows me what ETB is. ETB, anybody? Electronic Toothbrush. I go one second, we have a category called power toothbrushes. And she says, "But in Latin America we call it ETB." And I said, "Now you call it power toothbrushes." It's not my decision. It's not her decision. We have a product team that owns the product names, right? And it becomes less controversial because it's not me who's saying no. If you want a product to add it, go hear the product team, right? It's just the way it works because I don't have a hope of one day, close your ears, tying back creative to SKUs unless at least I can agree on what the product is called, right? You have to build that trust. And by the way, that holds true for campaign names and initiatives, right? It's not the buyer that decides the campaign name. It's the planner that decides the campaign name. Anybody want to know how many different ways you can spell the word traditional? Because there are a few.
I love that. This is an example I've been using for four years and the first year it really works. Is that if it's a back to school campaign, right? How do you write back to school? Is it back to school? Is it BTS? We love our abbreviations at Colgate, right? No. The planner is calling it back to school. The buyer gets to pull down a list that says back to school. There's no more free form in our taxonomy. It's all established upstream. Now on paper at the beginning, it was very controversial. But again, ultimately, I think when you hear the way I'm explaining it, it becomes a lot less controversial and at least you're not taking the blame.
So understanding data and taxonomy ownership is super important.
Creating a unified customized data dictionary. I've heard a number of times in a number of sessions today, normalizing those definitions. If I have what we would call, always on an always on campaign. Is that a campaign? Is that an initiative? Is that temporality? You just doesn't matter. But everybody just has to agree going in that this is the way we're going to do it, to get consistency.
We're now using Claravine as a workflow where the placements are built on the campaigns and the creatives are built on the creatives. It's a whole hierarchical approach and Claravine is enabled this to have a hierarchical approach so that we have an entire metadata base that sits in Snowflake of all of our metadata and all of our activations.
It's a very powerful tool because even if you're going to pull directly data from a platform, being able to have that lookup to be able to say, "Here's where the metadata mapping is a hugely powerful tool." This is not a set it and forget it process. You need to maintain data standards over time. We've started now loosening the reins a little bit...
In terms of who gets, I mean, literally when we relaunched 2025 our campaigns, and our placements, and our audience, there were three people who had access to make changes globally. Again, a billion dollars in media. Three people got to make changes to the dataset. We're opening it up a little bit more on a case-by-case basis. But ultimately the idea, as I said before, establish the rules upfront. What do I want to know from my brand plan all the way upfront? How am I going to chase my whitening moms down on a delivered tactical execution? You need to understand that process. In year one, when WPP signed on with Claravine, we were at 20% compliance on campaign names, which means 80% of our campaign names in the system needed to be corrected, right? I don't think that's a crazy number when you think about it, if you know how these systems work. By the end of year one, which is, let's say, 7 or 8 months, we finished the year at over 95% compliance. Meaning no exception reports, which then gave us the now the gumption to go ahead, okay, now it's not just campaign names, it's placements, it's audience, and now tackling creative. So it's a huge that foundation enabled us to have the confidence, better word, the confidence to go ahead and take the next step in our journey for better metadata standards. With good metadata, good reporting and analytics is easy. With bad metadata, it's impossible. And there's not a lot of middle ground.
We also, as a team, and again, it's us partnering primarily with our global team at WPP, and then our regional teams at WPP, we understood the areas in which we can compromise, whether it's specific taxonomies or use of specific fields, and also understood where we weren't going to compromise. You don't get to decide Canada's a part of the United States. I'm sorry. Go talk to the CFO. There are certain areas where we're just not willing to compromise. We've gotten to a point where globally we care a little bit less about specific taxonomy uses. As long as it comes through Claravine, we're able to do it. And by the way, I don't think we talked about this. Claravine has integrations built. So I can actually use Claravine to check everything sitting in Campaign Manager and make sure that's all compliant and people didn't bypass. So building in our flexibility in our process and knowing what our priorities were. And we've really come a long way.
We used to have multiple taxonomy. Whoever did the taxonomy would come up with their own way, even within teams. I could be looking at programmatic reports out of Europe, and the same team is doing it multiple ways. We're still here in our 2025 relaunch, which happened in November, with one single campaign name standard. I'm not sure that I still believe that. We have multiple different versions of placement name because of all of the different systems, but they all tie back to the same record. It was a huge change. The values are established, as I said, super important. Values need to be owned by the people who own the values, right? It almost seems simple, but it really isn't.
Everything should be linked where possible. Use it. Use a process flow to understand who does what first, who knows what when, and get it to be captured. You wouldn't expect the ultimate the person who's putting the trafficking instructions in for a placement to know that this ties back to a whitening mom audience. It's not something you would expect them to know. We did not have incomplete governance, and now we have a much more robust governance system in place which we continue to build on. So that's and we have our official review committee. Everything gets vetted and everything gets checked by a central team, right? That's the story. The truth is one of the most immediate results is you're not filling out exception reports every week. Or you're not chasing down people more accurately to fill out exception reports every week, right? It allows, as I started, it allows the data flow to flow without human intervention.
When you bring the analysts to the table, it really enables them to have a voice in the things that they want to analyze. And again, I think even with all of our constriction, we have a pretty agile global taxonomy.
You said I was always fast-- - Yeah, but-- - I think it was going to be-- So just a couple quick things before we wrap up and open to questions.
A couple key takeaways. Number one, this is something you cannot afford to ignore, especially with this explosion of GenAI and all of the exciting things that are about to happen. Just take a step back and think about your data standards, your process, the collaboration that happens with your partners. But as we think about what's next, Mark talked about the complexity with campaigns, he did touch on creative. I think that is one of the biggest opportunities. Data standards for creative, making sure you've got the right metadata, the right naming conventions, the right IDs in your DAM. And with the exciting announcements that Adobe made this week about content analytics, you're going to have to have your data in order to really get the value out of those tools. The number one thing, take the time to think about this because it's the old adage of garbage in garbage out, especially with AI and models. Make sure that you've got the data in the right format, in the right structure, you take the time to do that. You've got collaboration globally across all of the folks that touch the data, which seems like a daunting task and we'll get to that in a second. But this is going to be your engine. This is going to be the fuel to everything you do AI wise. And if you get a head start on that, it'll give you a leg up on the competition.
So finally, just key takeaways before we open up to questions. I cannot stress this enough. It's the commitment and the drive to get a global alignment, and Mark said something so vital. Get the right people to the table, make sure your partners, your agency partners are at the table, make sure the analysts are at the table, make sure that everybody that should care and touch the data standards are at the table to help drive this forward. Invest in a source of truth. So a lot of the folks that we talk to, they use spreadsheets, or they think that they've got this set and standard, but make sure that you do the diligence and you invest in a source of truth, a platform that can help you get that consistency, that collaboration, and that governance. - You want to touch on the last two? - Sure. As I said, you have to balance the evolution and the revolution. I think the revolution will come in at the end of the process in terms of what you can do with it. The evolution people, the change is hard. And again, I think some of the biggest change is the mindset of the practitioner that they don't get a chance to grade their own paper. I mean, that's a really big change an organization has to make. And also as part of the process, what we ended up doing is we sat down with multiple of our agency partners, but certainly mostly with WPP over the course of a couple days of workshops where we actually watched all of the handoffs of the whole process of the data. This is what I do in Prisma. This is what I do in Campaign Manager. This is what I do on Meta. Each of the teams came in so we could see what we had to capture in taxonomy and what we could rely on the applications themselves to capture as well, also realizing that there are a lot of global differences you're going to have across the process. And seamless measurement takes planning. There's a line that I coined when I started working a number of years ago on AT&T. I need to spend eight months so I can answer every question in eight minutes. The only way you're going to be able to hit a button and get the answer is up there, there's prep for that answer, and that takes planning, and there's no other way around it. There aren't a lot of shortcuts. Great.
All right, well, if you'd like to stop by, we're at Booth 371. Mark and I are going to be there all day today. I'll be there tomorrow. If you want to talk about Oregon Trail too, I'll be there. But thank you all, great questions, appreciate the time. Thank you.
[Music]
