Zeta Global Holdings Corp. (ZETA) Earnings Call Transcript & Summary

All right. I think we are ready to get started. Thank you, David, Chris, for coming. Really appreciate you guys taking time out of your busy schedules to be here.

It's our pleasure to be here. Thank you.

Yes. And with that, we can get right into it. So Marketing was one of the first applications of the Zeta platform. You've increasingly talked about the company more as this intelligent AI infrastructure asset. How do you think about what data can become over the next 5 years beyond the core Marketing Cloud?

That's a great question. Trying to keep my voice here. I've been talking -- you've had my talking since 8 AM this morning. When you think about the evolution of our business, you have to go back to some of its foundation. When we originally started the business, we started the business as a business intelligence company. We actually just focused on marketing because it was a $1 trillion TAM. And it's been a very, very positive thing for us. As we think about the evolution of our business it really revolves around intelligence. We think there's going to be 2 types of software companies. There are going to be software companies that create intelligence and have proprietary assets, and then they're going to be workflow management tools that they're going to have problems in the future. Because you've got to be able to take data, create it, synthesize it and then make it actionable. And that's really what I think we do best at data. When you look at our core marketing business, today, we returned between 6% and 700% return on every dollar spent through our platform in marketing. Now what we're seeing with the adoption of Athena and gave some updates for Q2, but we're seeing faster adoption than we expected, and we're seeing better return on investment than we expected. And when you think about Athena, I think Athena is really the future of our business. And not to get too esoteric, but if you go back, humans and sort of our current form existed for about 400,000 years. For almost all of those years, we've communicated via voice, right? And then you look at some of this enterprise software, most individuals use between 3% and 5% of the capabilities of their enterprise software. I always sort of joke with this type of an audience, what percentage your Bloomberg terminal do you actually really use, right? So Athena supersedes the creation of the keyboard of the 1950s, which I think has created meaningful friction between humans and technological adoption. I think we're now at 83% of people utilizing Athena or utilizing it from a voice perspective versus a tax perspective. And we're seeing, as I said, a substantially higher uptick in sort of adoption and a meaningful increase in return on marketing spend for enterprises that have adopted Athena. At Zeta Live, we're going to make 3 of the biggest announcements we've ever made. They will be the evolution of Athena to answer your original question, where we start helping enterprises to run their entire businesses, where it's not just customer acquisition, retention, monetization or business intelligence. It's the actual seamless integration and superior operating of their core businesses by using Athena and our data cloud.

Yes. I think one of the focuses of you has been getting more enterprise -- big, large enterprise wins. And I think one good example of that is Gap, Gap selected Zeta as part of their broader marketing stack transformation. So what's kind of changing in the market that's causing some of these large enterprises to reconsider their incumbent marketing platforms now?

Well, we -- first of all, they made us their system of record. So I'm not sure we're part of the ecosystem, I think we're the core of it. But what I would say is 2 things have really changed. One, AI has now become so prevalent, but most enterprises don't know how to adopt it in a way that creates return on investment. That's what Zeta does for a living. Because we're able to create such a high return on investment, enterprises that in the past might not have worked with us are now saying, yes. The other issue is we're in, I think, the largest sort of replacement cycle of existing marketing and technology stacks I've ever seen. And I thought it was peaking a year or 2 ago, and it's actually now accelerating. And it's accelerating at a time where I would tell you 5 years ago, nobody had any idea who we were. I started every meeting with Zeta who, meaning who are you and why are you here? We then moved a couple of years ago to sort of why Zeta. They know who we are. They know why we're in the room, why are we going to pick you. We've now crossed over to Zeta now. It's really become enterprises feel they need us to be a part of their tech stack where they're not going to keep up with their competitors. Now we'd like to get to must-have Zeta. That's sort of the holy grail for us. I'm not sure we'll ever get there. But that happening. Simultaneous to the opening of testing with AI for return on investment, simultaneously to the evolution of our brand, simultaneously to the massive upgrade cycle for technology and marketing clouds, I think, has been a driver, as you know, better than most. Last quarter, we grew the business 44% top line, 56% EBITDA growth, 73% free cash flow growth. And we all know that numbers are outputs. The input is the business working. And our business is really working right now. I actually -- I think it's working better than at any point we've ever operated the business.

Yes. And then data has been part of Zeta's differentiation for a long time. As customers increasingly bring their own data in through platforms such as Snowflake and Palantir. How do you think about the role of the Zeta's data moat, I would call it, evolving?

Shockingly, 2 of our biggest announcement this year, right? In the first 6 months of this year, we announced that we are amongst OpenAI's first enterprise partners, that Athena was going generally available, both for the enterprise and the agency. We announced our strategic partnership with Snowflake and probably the largest deal we've ever announced was the Palantir partnership. Both those companies are incredible companies. We've, in fact, re-architected all of our data cloud on top of foundry and we've adopted Palantir's ontology to help onboard data and streamline it between us and enterprises. But in no cases does our data go into their platform. They don't see it. It's the same thing with Snowflake. Unless we're using a clean room sort of a copy, which has been a massive win for us to be able to go in no copy and go joint [indiscernible] because it's accelerated the time to test. It's lowered the barriers to testing, and it's now got clients saying, yes, much, much faster. Our entire business based on enterprise is giving us their first-party data, but merging it with our data. and that's where the magic happens. So the fact that they're able to move their data faster through Snowflake, Palantir is a massive win for us because they're partners of ours. We don't expose our data even to the enterprise client. So when we combine the data, we remove the personally identifiable information and we replace it with Zeta ID number. That does a number of things. A, it allows us to import the 5,000 to 7,000 incremental data elements that our clients might not have on their end users, demographic, psychographic, credit card transactions, credit scores, all the way through. And not only does it allow our algorithms, which we've built internally, we built all of our own inference-based models. It allows the algorithm to get smarter, smarter, smarter, but if the enterprise were to ever fire us, they would lose 100% of that knowledge and never transfers to them. I think that's one of the reasons we had a net retention rate of 120% last year and why we continue to have a much higher net retention rate than I think we've projected to at [ $110 million ] to [ $115 million ].

Yes. We have to spend a little more time on AI. So Zeta offers prebuilt agents across campaign creation, audience building, analytics, quality assurance. Which agents are gaining the most meaningful production usage rather than just experimental engagement? And then how do you think about the value that those individual agents versus what Athena provides is kind of that aggregate...

Well, I was going to say #1 by far is Athena. Like we saw 40% of our superscale clients, right, adopt Athena -- in which and they're doing things that we've never seen before in the platform from a scale perspective. It's -- they're building audiences, they're looking at new technologies, they're looking at new activation capabilities at, I mean, rates we've never seen. So I think Athena is really going to be sort of the most powerful one. As it relates to other agents, they're all subagents that then sort of work together. So audience-building agents then lead to activation capability agents that then lead to agents that help you lower your churn. All of them sort of begin to compound and work together. The other thing that's been really cool from my vantage point, I'm a nerd, so the fact that I think it's really cool might not mean others think it's really cool. But the uptake of our build your own agent has been astronomical. So we're on our third edition of our AI agent studio where our clients can now use Athena to voice enable build their own agents, to build and sort of handle tasks that they otherwise would have had to do manually. You go back to the average enterprise user uses 3% to 5% of most enterprise software. The more agents they're building, the more they adopt Athena, the more they're able to use the entire capabilities of the platform and the data ecosystem. The better the return on marketing spend they're seeing, the more they spend with us, the higher our net retention rate. So it's sort of -- it's a nice funnel.

Yes. And then to go into that metric that you were talking about, 40% of super scaled customers have became monthly active users of Athena within like 4 months of launch. So how do you think about what's reduced the friction of adoption relative to maybe some of the other enterprise AI products that we've seen that are taking a bit longer?

I think Dave mentioned it's the voice enablement. And of those 40% active users, we -- what we know is that they're growing faster than the rest of the set of superscale customers. But -- and David made kind of a slight mention to it, but just to give you a sense for how our revenue funnel works. Customers pay Zeta on a license and a subscription basis to effectively use the data cloud to create automations around audiences and campaigns. Of those 40% monthly active users, 83% of the engagement is voice, and what we're finding is creating thousands more audiences and campaigns, which if you now think about downstream from that, roughly 50% to 60% of our revenue is that subscription to the data cloud and the platform. The other 40% is the monetization through consumption. That consumption happens once you create an audience and put a campaign workflow in place, then the software of Zeta helps you understand deterministically which individuals in that audience are most responsive to CTV, display video to mold to e-mail. Once that curriculum of workflow begins to operate that drives the meter of usage with them. So it's an early positive indicator in terms of the intent of the Athena platform is to create facilitation tool, more easy to use and access more of the platform and it appears to be working.

Yes. We're already seeing it flow through. But it's still early days, and I think we'll see it continue to flow through. I mean, listen, the business is really working right now across the board. When we set out to grow the business, we really look at it normally, like how do we get to half our revenue growth is from existing customers and half our revenue growth is from new customers. And it's certainly been much higher than we expected on both fronts. We're adding more customers than we expected to that are spending substantially more money than we originally expected them to do that, which is a good combination when you're running a business.

And what do you think it is specifically about voice that is kind of changing things for your customers? Like what are they not doing on the traditional interface that they're not doing with the voice interview?

Yes. So I think the big thing there is we didn't just create a voice interface to navigate the platform. We did -- we actually did that a few years ago. We built this platform called ZOE, I think, standard for...

Zeta Opportunity Engine. They're so good, the acronyms. We have so many acronyms, but I have...

[indiscernible] I forget. We have more acronyms now than I can keep track of. But ZOE was effectively a voice enablement to better just navigate the platform. Customers who adopt ZOE spent 250% more than customers that did not adopt ZOE and she was rudimentary at best. The big difference with Athena is, Athena has been built, so you just have to tell or the outcomes you want. You can literally tell Athena, I'd like to create 2 million incremental customers this quarter, and I'd like to lower my cost to create customers by 15% while doing it. She will navigate the entire platform, the entire data cloud and in real time, show you what you need to do to do that. You can then say, great Athena, let's start with test to $500,000. Could you e-mail me the reporting every hour on the hour because I'm going to be out of the office because I want to see the return on spend here. Activate now, and she will activate for you. She will then send you the e-mails every hour in the hour. I don't want to get too ahead of what we're announcing. But ultimately, we believe that you should be able to access Athena through any UI, not just through ours. So you'll be able to go into whatever platform you're using, I'm going to get in trouble for this later with the tech guys. But the reality is you're going to be able to talk to Athena and say, great, the test is working, let's go live. And that's where I think it's going to be very, very interesting.

And I know you guys love to talk about ROI. So how do you think about the incremental revenue opportunity you're getting from this higher usage that people are getting with Athena versus the cost that you have to absorb to support higher AI usage and kind of what it implies for Zeta's longer-term margins?

Well, I think it's interesting to note that last quarter, just -- it's not your exact question, but as you know, I'll answer the question I generally want to versus the one asking me. But I'll start with last quarter, 89.6% of all new code generated by Zeta was generated on an automated basis. And yet, to answer your question, less than 1% of our revenue, total is a company was spent on token utilization, well under 1%. So when you think about it, we've built workflow management tools, starting with Spade, rolling to Loom to allow for workflow management to the best foundation models at the lowest token utilization, while doing automated QA before it gets to an architect who then pushes a button and makes it generally available. our ability to build the vast majority of our AI internally. All of our inference models are homegrown and home built. So as we scale as a customer, our AI utilization expense has not gone up at all. In fact, I would say it's gone down as a percentage. Yes, more or less. Over the last few quarters, I think it will continue to go down. We were able recently to cut a very interesting deal with one of the foundation models that will allow us to scale very, very large at a fixed cost.

So you talked a little bit about this earlier, but Athena is kind of evolving from just answering questions to really executing actions and recommending things for your customers. Where are you seeing customers be more comfortable delegating decisions to Athena versus where are they still kind of a little hesitant and want more human in the loop?

I think that is a great question, which is -- not to say your other questions weren't great. But the reality is that we are seeing a lot of exploration with Athena. I mean, clients are really comfortable exploring in their data sets, in our data sets and the merge data sets, where we're not seeing everything yet is activate now. We're seeing them spend more time making sure that they're double and triple checking the numbers, the accuracy levels have been at the [ 59 ], to be clear. And we're even sending in some of our sort of FTEs to help clients really understand the accuracy level of the outputs from the explanation -- exploration, I should say. I think it's -- I equate it to when we all started using the Internet. You were probably a little young for this example, but I say that as a complement. You lost 90% of transactions on the credit card entry page. People were terrified to put their credit card into a transaction. I mean, I lived through this. I'm sure a lot of us had for many, many years. Today, you just, who cares, right? Put your credit card in, even if it's a site you've never been on, you just put it in and you know it's going to be fine. And if it's not, your bank will cover it and so on and so forth, although most of us go back to Amazon and just push the buy now. We're seeing a similar thing with Athena. She's getting them to that last page. They're looking at it. They're doing the activation, but they're double and triple quadruple checking it before they're going into the activation mode. I think that will continue to scale.