I don't know if my mic is automatically on.
No, you'll just hear it.
All right. I think we are ready to kick off. Thanks so much for joining us, Day Three Goldman Sachs Communicopia + Technology Conference . Delighted to have David and Chris from the Zeta Management Team. Thank you, gentlemen, for joining us.
Thank you.
Thank you for having us.
A lot happening in the industry. Perhaps we'll start, David, with a little bit of your vision for how you think the ecosystem is changing. One of the things that's really stood out to us is Zeta consolidating wallet share at your customers, both in the marketing technology sphere and in the adtech sphere.
Yep.
The question for you directly is, where are you consolidating? What do you think that tech stack, marketing tech plus adtech, what do you think that ecosystem looks like three to five years from now?
Thank you for having us. I think you're going to see a continued combination of what is called marketing technology and advertising technology because it doesn't make sense to operate those as separate tech stacks going forward. If you think about our platform today, we have three separate use cases: customer acquisition, customer retention, customer monetization, which are traditionally in different components of those stacks. You have 14+ channels which can be activated across. If you think about our business, which, as you know, has grown on a compounded growth rate of greater than 30% top line, greater than 50% bottom line, and in the mid-70s percent from a free cash flow perspective over the last three years, you've seen us grow largely by going from an average of 1.2 channels per client to an average of approximately three channels.
When you think about the step function for our business, on a higher level, most companies like to talk about their total addressable market. I like to talk about the fact that my 567 global enterprise clients will spend $100 billion this year on marketing. I don't just look at the trillion-dollar marketing TAM, I'm looking at my existing customers. At the middle of our range, we'll have about 125 basis points of wallet share, which is up from 100 basis points last year, which is good. When you think about the business, we started looking inside of it. Growing from 1.2 channels to three channels has allowed us to grow very nicely. We saw that clients that used us for multiple use cases simultaneously had by far our highest NPS score and spent, excuse me, materially more money than clients who used one use case.
Third, they had by far the highest return on investment. When you think about Zeta 's ability to get from 125 basis points to what would be 200 basis points approximately in our 2028 plan to what I believe we can do, which is 500- 1,000 basis points of wallet share, how do we build a $5 billion or $10 billion a year revenue business? The way we do it is with a new strategy called One Zeta. One Zeta is very simple. It's convincing the vast majority of our clients to use us for multiple use cases. Interestingly enough, the more use cases they use, the higher the return on investment to them. It makes sense, right? Because if you're doing their acquisition, that information informs retention. That information informs monetization, which in turn both inform acquisition.
The CDP that's utilizing the artificial intelligence is getting smarter and smarter at all three simultaneously by having each one in the individual stack. By the way, that's why we hired Ed See . He was the Head Partner at McKinsey for running their Chief Marketing Officer practice. We were able to get Ed to leave McKinsey and come become our Chief Growth Officer. His real job is, we call him the One Zeta guy.
Yeah, absolutely. Tell us a little bit about what that means for that $100 billion in budget spend. Is there a way to think about what are the big categories within that? Where do you already have the right to win? Where does someone like, I think about Meta talking about reevaluating their ad program and how much more analytics they're putting into that, how does that fit into where you fit in?
Yeah, I mean, two different questions. We have a very unique and very good relationship with Meta. In fact, many years ago, our company built their API for them. We have a matching between the Zeta ID and the Meta ID. Meta is embarking on a project to automate the way their marketing works. That's primarily because the vast majority of their customers are small to mid-sized businesses, and they do an incredible job there. For large agencies and large enterprises, the data match is going to be very, very difficult. Our data cloud, which informs every single campaign that's run into Meta, cannot be replicated outside of what we're doing. We actually believe we'll continue to grow there and continue to do very well there. We have a very good relationship with Meta.
I can't tell you exactly how much we pay them, but the vast majority of the revenue that flows through us goes to them as it relates to that integrated platform, not the direct platform.
Sure.
As it relates to where we think we can go as a business, today, 50% approximately of our client spend is digital, which is addressable to us today. The other 50%, I believe, is going to become addressable. I think in the next five to 10 years, 100% of marketing in the United States of America is addressable.
Are you saying that it will become?
I'm saying it will be digital in some way, shape, or form.
In one thing, OK.
I believe the set-top box in the house, even where it does still exist, will become addressable down to the IP address. We're already working on solutions for that today.
OK.
I believe that even over-the-air radio, you'll be able to build attribution models when you have access to the transactions for capabilities out of the credit card processors that can be processed by vendor, by SKU in the geography in which you're playing. We're already working on stuff around those two ideas. Today, 50% is digital. I think that goes to 75% over the next five to 10 years. I think a lot of that is connected TV. If you think about your own lives, how often are you flipping on the cable box nowadays versus watching YouTube television, Hulu TV, Netflix?
Targeting is terrible.
It's terrible. In fact, I laugh because my wife subjected me to The Handmaid's Tale, which was, you know, it was good the first season or two. It got a little.
You watched it with ads?
Yeah. Oh, yeah, I'm cheap. I watched it with ads. I'm the weird dude, Gabriela, who likes the ads. I'm watching them full.
Intelligence. I got you.
You know, it was funny. I couldn't tell you how bad the ads were. I saw the same ad because you watch the show like three or four times in a row.
Yeah, you see the same ad 12 times.
The same ad 12 times. Our platform already builds in frequency caps, and our platform is already building targeting. It was great because one of our largest global customers is an automotive insurance platform, and I saw their ad targeted at me twice in the four shows. I was very pleased about that. I think it's a very big opportunity, and it's our fastest growing business.
Yeah, fantastic. OK, I want to ask you the AI question very specifically.
What's AI?
I'm going to ask not that question.
That would be dope, by the way. We're an AI company.
If you were to look back over the last 12 months of technology milestones that you've hit with your product team, what is the one milestone that you're most proud of?
Let me start by saying we started programming in artificial intelligence in 2017. When we went public in 2021, the banner we put on the side of the New York Stock Exchange said, "Data plus AI equals intent." She's laughing because she's heard this joke. Everybody said, "Who's Al? And why is he in charge of your AI and your data process?" We've been doing this for a very long time. What I would say, the thing I'm most proud of, we're going to be announcing at Zeta Live in a few weeks. I won't point that out because I can't talk about it just yet. Other than that, we have already launched the third iteration of our AI Agent Studio. We have strung together today three AI agents into one agentic workflow. Now try saying that 3x fast.
In all seriousness, we have one agent today that is doing all of the targeting and looking at trillions of data points. Simultaneously, the next agent is figuring out the best place to target that exact individual. Third, and probably most interesting from an intellectual perspective, is a real-time attribution agent informing the first two. I actually thought that 1 + 1 = 4 . It actually is 1 + 1 is an order of magnitude smarter. 1 + 1 + 1 is two orders of magnitude smarter. 100x more intelligent.
Critical.
Yeah. By the way, the return on investment to our clients that have adopted this agentic workflow is game changing.
Amazing. (Inaudible) , over to you.
I want to ask the death of SaaS question. It can come up a lot with investors over the past couple of months. What's your kind of view on this hypothesis that investors are coming up with that we're just going to have these LLMs sitting on databases that do everything for you? How do you view your tech stack as continuing to be differentiated going forward?
I'm going to let Chris talk about how our consumption works and subscription plus consumption. Let me talk at a very high level. Software is going to change. There's no question about it. Vibe coding is going to change the playing field in the way organizations can adapt and create software both internally for themselves and external companies that provide these as services to other organizations. Let me start by saying that the single most important thing that gets fed into any algorithm is data. Data is the lifeblood of large language models, small language models, and what I lovingly call what we do, which is mid-sized language models.
Our proprietary data cloud of 550 million people globally and the 5.2 million publishers who have embedded us as a first-party component of their technology stack and the ability through our JavaScript on-page and our first-party tracking pixel on-page to ingest information in ways that others have no capabilities is a moat around our business that will not only survive through the changes in AI, it will thrive meaningfully. As it relates to other legacy SaaS-based companies, I don't think anybody's going away this week or next week. What I would say is it is almost impossible to integrate AI as native to a legacy architecture. A lot of enterprises are trying to do that. You might have a platform, but the platform has to step out to an algorithm to do a query.
The algorithm needs to do a data dip and then back to the algorithm to create intelligence and then inform the platform of an action. There are some industries that doesn't matter. In our industry, that destroys return on investment where we need to make a decision in a millisecond. Back to 2017, we decided to eliminate our legacy marketing platform. We ran it for cash flow for a few years. We architected an entirely new platform that we rolled out in 2021 that put AI and data as native to the application layer. By controlling the CDP layer, the platform layer, and the data layer, you're able to decide in a millisecond question. I'll let you touch base on the consumption versus.
Yeah, I think what is sometimes missed externally that our customers see every day is this blend that we have of the software recurring part of our revenue model and the consumption part of the revenue model. The data cloud that David described, the intelligence that sits behind it down to the person level, is what is doing the recommendations for which channel each of those individuals distinctly should be served an advertisement to based upon what they are most responsive to, how they like to engage. It's the software on the front end that is determining what are the best channels to go to. If you think about our revenue model in that context, about half of our revenue is being performed by the algorithms deciding which audience should be put in place, how they should be orchestrated, and where they should be pointed towards.
The consumption part of our revenue model is that point of impact for our customer, which is the activation. Do they get served a combination of an email, a CTV ad, or display video, or some combination of the dozen plus channels that we have? The revenue model has stayed pretty evenly distributed over the last several years. As we've grown an average of 30% over the last four years, high 20s percent organic, the revenue model has stayed about 50% consumption, 50% recurring.
Yeah, and Chris, maybe just on the conversation that we were having a couple of weeks ago, I would love to hear you talk a little bit about how Zeta's pricing model could evolve. You're already so closely tied to value. Is there a way that actually you can even get closer to value-based outcomes?
Yeah, look, we want to create an environment for our customers where there are no inhibitors to doing more with us. As David has said before, I've said before, we like being the low-cost provider because we can get in at a very good, attractive place and then grow our wallet share over time. How do you create that ease to do business with you? Today, that revenue model is tied to X millions, tens of millions of emails sent at some unit price, X impressions made at some unit price, X mobile messages deployed at some unit price. Can that evolve to a credit-based pricing model, much like a Snowflake would have? To your point, because we sell and get better with our customers tied to the ROI that we generate that's attributable and verifiable, how do we participate in that value creation? It's going to be a partnership.
It's not going to be something that we turn a light switch on. We're doing a study with one of the premier, if not the premier, consulting practices that I've incidentally used in the past for a similar exercise with great outcomes. I think we're going to be informed. We're going to be very measured and disciplined. It's not priced in any of our long-term models. If we do something there interesting, it'll be upside. Yeah, excited about what we'll learn.
By the way, in our 2028 plan, we're already projecting a 500 basis point increased operating margin with a meaningful increase in percentage of free cash flow. That is not priced into that. I do think there's an opportunity specific to some of the channels we operate at where we've really become the dominant player, where there is pricing room.
You spoke a bit about the importance of data. How do you think about continuing to have relevant data out into the future, particularly when the way that consumers interact with the technology ecosystem is probably going to change?
Yeah, that's a good question. It's interesting. I think people don't realize that the 242 million active Americans who are opted into our data cloud and the 550 million people globally, that's a one-quarter number. If they haven't interacted with us or opted in in the last quarter, we move them to an archive status. We've increased the size of the data cloud every year for 10 years in a row. I feel very confident we're going to continue to maintain those rates and continue to grow them. As I've said, we have 5.2 million publishers who are fully integrated into the platform, and not one of them makes up even close to 1% of the data in the data cloud. I think for other companies, it's going to get harder and harder to build high-quality data ecosystems where you're going to have the walled gardens continuing to consolidate.
I think some very large publishers are going to start to collectively work together in ways that will allow them to compete. Because, you know, let's be frank, OpenAI has opened up the ability to effectively ingest the Internet. Google has rushed through that door, as I would do if I were them. If you look at Gemini, which is good and getting better, the percentage of questions that are being asked and answered on platform is skyrocketing versus the percentage of questions that were being asked on platform and sent to third parties, whether it's e-commerce or whether it's publishers. Publishers are now sitting out there trying to figure out how they are going to restore traffic to their platforms, which, by the way, is a business we're in.
Quite frankly, it's been getting a lot of attention from our clients as of late because they're trying to figure out what to do. The publishers control a very large amount of data, and I think they're going to continue to control the data. I think their sources of traffic are going to change dramatically in an LLM world where, particularly, I think Google, I mean, there's no question in my mind that OpenAI is going to be rolling out an ad network. We'll partner with that just like we partner with every other network that's out there. It's just going to continue to eat away at the traffic that publishers see, and they're going to have to combat that. I think they're focused on it.
Yeah, and on the channel side of things, you guys allow customers to use a very wide variety of channels. That's not been the strategy of every vendor in this space. I'd love to hear about how that adds to your competitive differentiation and how you view channel expansion over time and continuing to meet consumers where they're at for customers.
Great question. I think most of our industry has been built as really interesting products that are businesses, whether it's a DSP or it's an ESP or it's a workflow management tool. We came at this in a very different way. We came at this that really what matters is the data and the intelligence. The activation methodology to me, quite frankly, is secondary. The fact that we own a DSP that's growing very, very rapidly while some other DSPs are having challenges probably is because we don't look at it as a pure-play DSP. We look at it as a fully integrated data ecosystem with AI-driven targeting inside of a programmatic platform. Same thing inside of a connected TV platform, same thing inside of an ESP, so on and so forth.
There have been challenges that other organizations are having, quite frankly, that we're seeing the inverse of those challenges. We're seeing it as tailwinds, as large agencies and enterprises want to have more efficient marketing and do it across channel instead of just in one channel. I think we've seen the benefits of that. Once again, we started three or four years ago with an average of 1.2 channels per client. Now we're at just about three channels per client. I don't see any reason we can't be at four, five, or six channels per client.
Yeah, it's a great point. Our last quarter, we talked about the fastest growing set of customers of Zeta 's in the second quarter were those that were using four or more channels, and that was up 40% year- over- year. It used to be, even internally, we would look at that three or more mark as like, are we continuing to make progress? It's interesting to see kind of the next leap now be those that are starting to knock on the door of four or 4+ .
They integrate, right? If somebody sees an ad on the Internet and they click on it and don't buy, we want to show them a connected TV ad that evening. If they see the connected TV ad and they don't buy again, I want to target them in Meta the next day. When they click on the Meta ad and purchase, our attribution platform ties all of those actions in the journey to one return on investment. They might do eight things before they purchase. The average platform that is a channel-specific platform, whether it's a DSP or a mobile platform, whatever it is, is going to say that last click was the greatest click in history. We're actually saying the journey was these seven actions before they purchase. By the way, these two were inefficient. These four were very efficient. We need to double down there.
If you have a one-channel strategy, how do you do that? We're a DSP. The only thing you should be doing is programmatic. That's it. We believe that social, mobile messaging, programmatic, connected TV, online video, and search are all very important components of long-term marketing and customer acquisition.
You brought up agencies. It's been a big part of your growth algorithm over the past couple of years. Can you speak a bit about the different levers you're seeing for growth in the agency business? There's a couple of different pieces to that. Also, can you touch on a little bit of what the financial implications of that have been? I know free cash flow has been a big debate. Just how you look at that normally.
I don't know if it's that big. We had a 60% free cash flow conversion rate last quarter. I'll let Chris talk about the second part. I'll talk about the first part. When a DSP goes into an agency, the first thing you've got to look at is the average DSP is charging a 7% exhaust rate, just 7% of revenue. Yet the publicly traded DSPs show a 19%- 20% take rate. How is that possible? Mathematically, it doesn't really make sense. They charge an average of 25% for data, and they're selling third-party data. They keep about half the revenue. You take the 7% + 12.5%, you end up at 19.5%. When we go into that same agency, Holdco, we give them the data for free. We own the data, so they can do one of two things. Now remember, the agency is making 8% approx.
You know, they've got creative fees. They're actually very good businesses, but on average for marketing spend, they make 8%. If we're giving them the data for free, they're now making 34% when they work through us. They access first-party deterministic data instead of third-party resold data based on a cookie, which tends to be inefficient. Everything we're doing is based on a first-party tracking pixel. Quite frankly, most of our agency clients are giving a big percentage of that money back to their clients and making them happier. They're increasing their margins meaningfully, and they're getting a superior product with a superior data set. I think that's one of the reasons we have seen our agency Holdco business growing so rapidly. I always joke, how do you win in the marketplace? You sell a superior product to your competitors at a lower price.
This is an example of doing that. Chris, do you want to touch on the?
Yeah, (Inaudible), to just to kind of frame the opportunity we have, we today work with the five largest agency Holdcos. We embarked on that strategy two, two and a half years ago. They've grown rapidly, and they're in such early days that they have a great continued rapid growth runway ahead of them. We also, within the last year, started to create that sales team that broke into the large agency Holdcos, stand up an independent agency sales team, which haven't yet been acquired, right? They could be very industry-focused. They're more niche-focused, but yet they deploy the same amount of media and advertising that large enterprises do, right? They're large in themselves. There's about 1,000 of them. Realistically, there's probably 200- 400 we could be working with. You can count on two hands, although it's been growing rapidly, where we're at today with those independents.
On the cash flow side, as you pointed out, what's interesting about, and this is all public, you can see this in their public filings, our enterprise direct relationships pay us in, call it, 55- 58 days. Our payables are right around 58, 62 days. There's no working capital disconnect on that side of our business. However, the agencies, it's just their payment model. They pay their vendors in north of 100 days. As that part of our business has been growing so rapidly, it's created a non-operational timing element where we have this deficit on working capital. Instead of having, call it, free cash flow conversion from EBITDA to cash of high 50%- low 60%, last quarter, if you just normalize, and you say this should have been neutral, which it will be eventually, we would have had cash conversion north of 80%.
It is a dynamic that I think is something we just need to double-click on sometimes with investors and make sure it's understood. It's probably our best use of our balance sheet, frankly.
Right. I was going to say, how do you better use your balance sheet than financing payables for some of the world's largest companies that pay you at a 100% rate?
Makes sense. One last one from me, and then I'll turn it back to Gabriela. You've given some really impressive ROI metrics. I think the ones are 50% of cost savings for customers when they use Zeta. Also, every dollar spent on Zeta, you see that translate to $5- $7 in revenue. How are you able to do that?
We do it by extremely good targeting, right? If you think about it, once again, we build a consumer data platform for our enterprise clients. We import all their data. We merge our data with their data. We eliminate the personally identifiable information to protect the consumer. We replace it with a Zeta ID number, so Zeta ID number 135789. We then merge the data from our data cloud into them. We match it greater than 90%. We're able to import between 5,000 and 7,000 incremental data elements: behavioral, transactional, psychographic, demographic. What are they reading? What are they researching? What are they searching on the web? All of that. An AI agent starts looking at all of that. Now, we're doing two things that I don't know of anybody else who can do them.
The first thing we do is when we do the marketing, we remove every existing customer that they have. Think about how often you get ads, whether it's television, electronically, whatever it is, for products you already own and you're not going to buy. We take all of that cost out of the marketing funnel. Next, the agent says, OK, these 3 million people are actively in market for your products and services, but this 1 million of them will not be credit approved by your criteria. We take them out. The agent goes into the market. The second agent is figuring out the most cost-efficient and the highest return on investment to target that person. The third agent is telling them in real- time what's working and what's not.
The ability to hyper-target by controlling the data elements, the ability to get focused on the best place to target based on return on investment to the client is how we drive that meaningfully higher return on investment. People will often say to me, how is your net retention rate 114% for last year? If they fire us, they lose 100% of the learning in that CDP. All of it goes away. I think that's one of the reasons we have such a sticky relationship with our clients.
What is the difference between a cookie and the first-party tracking pixel? What is the risk that the regulatory environment changes on the latter?
The third-party cookie, which we had hoped would go away, is not going anywhere. That's unfortunate. We would have liked that. The rest of the industry wouldn't have. It certainly hasn't hurt us. We grew 36% last quarter. A third-party cookie is you are working as a third- party inside of another platform to get some information on an individual. It's very difficult to tie it together, and it's not deterministic. It's probabilistic. All you know is that this cookie number, that person did these three things on that site or getting to that site. A first-party tracking pixel is a component of the first-party tech stack of the publisher, meaning we know not only what that person's doing, we identify who they are deterministically. By way of example, I can figure out if they can afford a product. I can figure out their credit approval rating.
I can figure out what they did on every site leading up to getting to that site. I can see often what they've searched as a part of making a decision on this person seeing an ad for one of my clients. A third-party cookie does not allow you to do that.
One of the really unique things about Zeta is where you sit in the consumer landscape. I know you have some really interesting data and some really interesting observations that you publish. Give us a little bit of a sense on what you're seeing for the durability of the consumer spending environment in the United States. What do you think is driving that? How is it possible that we went through so much volatility, so much debate around pricing, inflation, and yet we're sitting here and the consumer environment's actually been incredibly strong? We'd love to hear your thoughts.
Yeah, so I mean, we do publish the Zeta Economic Index. I want to be very clear that the talk of whether the index goes up or down and its effect on our business is ludicrous. We put it out for information purposes. If the score goes down, it has no effect whatsoever on our business. It's just how the consumers are acting in the United States in their entirety because we're looking at all their behaviors. We get every credit card transaction for every consumer in the United States. We compile it into that. I think you're seeing a bifurcation of individual if you look inside the numbers. The wealthiest 10% of America is spending more than ever. The bottom third of America is spending as little as they ever have. The middle continues to get squeezed to the bottom or the top.
Interestingly enough, the squeezing of the middle class in the United States is not just going down. We're seeing people in the middle class get wealthier because the markets are performing well and people are doing well. Now, probably I don't want to get into any of the percentages. The reality is that because the markets are performing well, people have money. Housing values are at all-time highs. When housing values are very high and when markets go up, people feel wealthy. People spend money when they're wealthy. We're seeing that. We're seeing, you know, they're not buying luxury goods, so to speak. They're buying a lot of stuff. Some of that might be a little bit of pull-through, you know, because of tariffs. I think largely the tariffs are being absorbed, which I would say surprised me. I didn't think that would happen.
We saw shocking when consumer prices came down last month. It came out this morning, which surprised me. I'm an economist by training, which is sort of pathetic. You can argue both ways for as long as you want. What I would say, though, is that the consumer is holding up. I think they're going to continue to hold up based on everything we're seeing.
Yeah, fascinating. Really insightful comments. I think the macro commentary is far from pathetic. Thank you for sharing that with us. Please join me in thanking David and Chris for their time.
Thank you.
Thank you. Thank you, Gabriela. Thank you, (Inaudible). I appreciate it.
She's off more than 1%.
I've had some good nuggets of oysters.
The funny thing is he does these 90% of the time without you. I only came to this for you. I do very few of these. He does a better job.