Palo Alto studio connecting Silicon Valley and Wall Street. I'm John Furrier, host of the Cube here with Dave Vellante, my co-host.
Hello, I'm John Furrier, your host of the Cube. We're here at the Cube's NYSE studio, of course. We have a Palo Alto studio connecting Silicon Valley to Wall Street. This is part of our NYSE Wired program and our open community. This is our mixture of experts series where we talk to the leaders and the experts in the field. Find out what they're working on because the AI revolution continues to power away. AI infrastructure is accelerating the software layers and in the system architecture and the stacks around building AI native applications. David Steinberg's here, co-founder and CEO of Zeta Global. Great to have you on, David.
Good to see you. Thanks for having me. John, great to see you. Thanks for having me.
The chairman, I'll get that in there because I wanted to get that in there. You're the chairman and CEO.
This is an interesting time. I want you to explain what you work on because you guys, I think, have cracked the code on what we've been covering on the data cloud. I call it modern legacy. You know, I have a cloud, I got stacks, I got multiple silos, I'm trying to stitch them together. That was kind of the SaaS generation, but now the new architecture of AI native looks a lot different, but the functionality is changing because you have essentially prompts, you now have applications embedding in MCP servers. So, you got a gentle coming in. So, this is changing the work. But ultimately, as the intelligence feeds the agents, explain what you do because I think you really nailed this.
I mean, I think when you think about what we are, we're an intelligent AI infrastructure company, John. We literally have one of the world's largest proprietary data clouds. 550 plus million global individuals have opted in to be in our data cloud. We have over 5 to 7,000 data elements per person. And our technology powers 5.2 million publishers. So, we're ingesting trillions of signals on a monthly basis that we then use solely to power our algorithms and train our algorithms. So, unlike a lot of other companies, in 2017, we made the decision to sunset our old architecture, re-architect an entirely new platform that put artificial intelligence and data as native to the application layer. That creates a speed. By way of example, we're able to create a decision in a millisecond with 10x the amount of data that one of our competitors needs 7 or 8 seconds to use 1/10 the data.
When you think about that, that's what creates massive return on investment, whether it's in marketing or whether it's in business intelligence, which are our two main use cases.
You know, it's interesting. You know, there's so many things going on in your world and our worlds together because digital and physical are coming together. Old school ad tech, old school martech stacks that have been around during the cloud era and like all the, you know, the email marketing era has changed. You now have completely disaggregated communities. You have massive amounts of data and signals. But if harvested properly, the specialism that comes out of that. And now people want low latency. Now, one more thing and then I'd love to get your reaction is that humans are great users of intelligence systems. How do I inform my business? But now you got agents. They're doing work. It's not just a search query. We're talking about like decision making. So, you have decision making intelligence going on here with multiple dimensions, which explain why this is important and why you.
Well, it's not just agents strung together doing tasks. We have four agents strung together doing tasks. So, one might look at what's the best possible audience. The second will figure out where to target at. The third is will that audience be credit approved by our client standards? Fourth, the attribution capabilities to build full return on investment as you're thinking about marketing. Every agent you add to a string, it's not 1 + 1 = 2. It's an order of magnitude smarter per agent. So, two is an order of magnitude 10 times better. Three is 100 times better. Four is 1,000 times better.
So, if you think of how a human would operate, when you start stringing three or four agents into a task, you're literally 1,000 times smarter. And when you're looking at an organization like ours, which is powering today 51% of the Fortune 100 and, you know, over 800 global brands trust us today to run their marketing CRM or business intelligence, when you start to look at the old-school marketing clouds, they're really antiquated, right? They've never redone their architecture to put AI as native. They keep signing new partnerships because the old partnerships are not working. So, they let's choose a new one.
Right. So, this guy's this hot this week, let's go with them. You're stepping out of the platform and, you know, there's a big press release. We're working with these guys now. But, the truth of the matter is unless it's native to the architecture, it can't actually create the functionality necessary and the speed necessary to create the true return on investment.
Talk about the technology and the business model issues because, I mean, you're talking about a couple things. I want to just unpack that. One, there's a technology play. You have a moat that you've created. We'll get into that in a second. But, your customer, they have data and they want to leverage that data, unlock the value of that data. But, in the old business model, it was I would just hire someone, an agency, an analyst firm.
Actually have a body shop.
Okay, give me intelligence so I can inform my data and so my people can make good decisions. Talk about that business model, because you brought up a good point. I got to hire that new firm. A lot of these firms are either old firms or new firms, but their mechanisms are different. So, there's like a one, a technology integration problem. And then two, the business models like, 'Wait a minute. This is a lot of loops, a lot of speed. I can't close the loops.'
One of the biggest value adds we do for our clients is we automate the sort of ingestion of all of their data. So, we might be with a Fortune 500 company. We're pulling their IVR data, their CRM data, their point of sale data, their prospect data. They've never had it in one place before. And we pull it into a consumer data platform. Now, a few months ago we announced the biggest partnership in our corporate history with Palantir. And what they're doing with their Foundry platform and their ontology is game-changing for us. So, we re-architected our data cloud on top of Foundry, and it's now allowing us to ingest data even faster, streamlining it. In the old days, you would effectively have a body shop come in.
Build it out from scratch, long cycle times.
John, recently we just had a massive win, and you look at that win, it was in a business intelligence use case. We competed with one of the three most important professional service firms. We weren't competing with a marketing cloud. And our data, which informed the BI, was smarter than this firm that is commensurate with, you know, the smartest people in the world.
All right, I have to ask you because you're getting, I mean, I want to get into the journey, because you didn't just get here overnight. You have a lot of history and get that data. And then now you're leveraging it. But in the analytics world, in the old days, 7 years ago, you'd have dashboards that told you what happened, not what's happening. So, there's now there's a trend towards what's happening, and then using AI to figure out what will happen.
And that's what we're doing. So, we look at most business intelligence is antiquated. It's literally you take 100 people, you put all the data in, you get a static chart and a graph, and you're looking in the rearview mirror. As Ted Lasso likes to say, there's a reason the viewfinder in front of you is bigger than the very small rearview mirror. What Zeta's doing is Zeta's using our data to build dynamic business intelligence to tell our clients what's next. So, we have a hotel client, we literally helped them figure out where to put their next 500 hotels. We have a client who's in the restaurant business. Where are they going to put their next locations? We have a client who's a massive retailer. How did we help them shift inventory based on online research for what people are going to be buying in the northeast versus the center of the country versus the west coast. They had their biggest back-to-school ever because we were able to help them.
So, your infrastructure, you're an infrastructure provider.
Yeah, at the end of the day, and we call it whatever you want, but to simplify it, because you're actually enabling, you're removing middlemen in the process.
That's right. We displaced a lot of people that don't like you.
That's right. When we recently, The Gap announced that they had chosen us as the system of record as their company, and they're obviously one of the most legendary retailers and fashion icons and out front, yes, we displaced Salesforce there, who had had that account probably forever, but we also displaced three other vendors that were in there, and four vendors became one overnight.
Actually, you don't really hear the vendor consolidation story in AI. Usually, it's vendor expansion. You're actually hitting both. So, talk about the value. Let's take The Gap example, and maybe some other customers can illustrate this, but okay. Vendor consolidation is almost like cost savings with AI. You get it, but the real value is that either top-line revenue or product service excellence. What is Gap getting now with you?
I can't talk specific to Gap, but let's talk about our average client. Our average client is getting a 283% return on technology spend with Zeta. And for every dollar they spend with us in marketing, we return between 600 and 700% return on marketing spend. That's a certified number by Forrester. That's not us just saying that. So, when an enterprise chooses us, they're winning on technology return on investment. They're generally able to reduce their head count pretty meaningfully, and we're able to return 600 to 700% return on marketing spend. By the way, I think this is one of the reasons, John, last year our net retention rate as a company was 120%. Our clients really love us. I mean, last quarter we grew the business 44% top-line, 56% EBITDA growth, 73% free cash flow growth, and people keep saying, 'How are you doing that?' It's well, we drive better return on investment. What we've done, you've been running a business for years, the only way to win is to drive value to your clients.
And they should get extract value. And you get and value is through profit. Happy customers buying more, more profit. All right, so let's get into some of the tech stuff. I know the numbers are great. By the way, you should be on our CMO series, too, because CMOs always tell us they're looking for modern solutions, and all the way, the shelf life of a CMO these days is almost 4 years. Brian and I were just talking about that. But we think they're more like CIOs. Because they're starting to get into more performance because they're used to performance marketing. Every CMO knows what performance marketing is, but they don't know what the modern version is.
Here's what they want. They want outcomes. And what Zeta guarantees is outcomes.
All right, so I'm a customer, so I'm a CMO. David, you got me. You sold me on the idea. Okay, let's merge our data together. But my other guy is over here in my IT department saying we should build a single tenant cloud. But because I'm giving you my crown jewels, so you're a cloud. Can you give me a single tenant experience?
Yeah, let me be clear. We don't take any of our clients' data. We create a consumer data platform. We input all of their data. They still own it. They have total data sovereignty. In fact, whenever Alex Karp and I are talking, this is like data sovereignty central. We then merge our data into that CDP.
And it's their infrastructure.
It's our infrastructure, but the ownership of the data, their data is theirs. The ownership of our data is ours.
And you're running on your cloud.
Correct. But we never use any of it to inform any of our other customers, and we never make our data cloud smarter from their data.
I wanted to bring this up because one of the biggest trends in our AI factories in the neo clouds is, and this is more of an Amazon disruption conversation, is that you have multi-tenant clouds essentially using AI to create essentially single tenant experiences because the workloads are end-to-end.
That's right. And Amazon has VPC, so like this has been around. It's not a new concept, but it does open up the efficiency side of having the scale.
What it also does is it creates zero data exhaust. So no foundational model has ever seen the Zeta data cloud. Nor does Palantir. It sits on top of their architecture, but they don't see the data. When we import our clients' data, our data merges it. They have their own algorithms that get smarter and smarter. Once again, here's what's really interesting. If they were ever to fire us, and I will tell you I can't remember the last time we actually got fired.
