Zeta Global has made 19 acquisitions in as many years. But, according to CEO and Founder David Steinberg, it turns down more deals than it closes.
“We’ll look at, I don’t know, 100-plus deals to do just one,” Steinberg said the week after announcing Zeta’s acquisition of two companies in two days.
On Monday, Oct. 5, Zeta announced its plan to buy Senso, a startup that helps brands show up in AI answers. By Tuesday, it had a second deal to announce: Digital Audience, a Netherlands-based data infrastructure company.
According to Steinberg, every potential M&A target is measured against the same five criteria: its ability to fully integrate within 12 months, customers who want each other’s products, employees who stay, a deal that pays off right away and technology or data that Zeta lacks.
“You’d be surprised at how many businesses only hit four of those and so we pass,” Steinberg said. “We’re very disciplined about it.”
Even so, the roll call is long, everything from comments platform Disqus, Sizmek’s DSP and Rocket Fuel’s DMP to digital survey tech platform Apptness, email marketing and identity company LiveIntent and mar tech company Marigold’s enterprise software business.
“I forget how many companies we buy till somebody lists them all,” Steinberg joked.
So how does it all fit together? Steinberg spoke with AdExchanger.
AdExchanger: Zeta was already working with Senso to power its GEO product, and Digital Audience was already integrated with your platform. So why buy them instead of just continuing to partner?
DAVID STEINBERG: We see Senso as much bigger than GEO. It’s how we influence the AI agents that will increasingly act as personal shoppers by making sure they have the right information about our clients and their products.
But we never share our Data Cloud data with any third party or feed it into any frontier models, so we could never give Senso access to it as a partner. But now Senso can use the full breadth of our data cloud to train models, and their engineers will sit inside of our AI lab.
Digital Audience is similar. The HTTP layer is dissipating, so having another tool for building deterministic identity is valuable. It also strengthens our European presence.
Neither company generates much revenue, but they punch above their weight.
You’ve bought a lot of companies, many of them before the HTTP layer started dissipating, as you put it. How do those older assets fit into the world you’re describing now?
The business has evolved meaningfully since the first of those deals. Most marketing companies are point solutions. We’re trying to deliver everything a marketer needs in one user interface and reporting infrastructure: a data cloud, a marketing cloud and more than 16 activation channels.
We’re channel agnostic. We don’t care how a campaign activates as long as it reaches a deterministic, targeted individual.
What would you say to the skeptic who just sees a pile of companies?
Don’t buy our stock.
Kidding.
Look, we break out our organic versus inorganic growth. I don’t know any other company that does that. Last quarter, we grew about 40% on the top line, 28% of it organic and 12% from Marigold. We do that so shareholders and potential shareholders don’t think we’re playing games with a lot of separate, disparate businesses.
We merge them and fully integrate them. The brands are sunset the day we buy the business.
When we spoke a couple of years ago, Zeta was calling itself an “AI-powered marketing cloud.” Now, you’re calling yourself an “intelligent AI infrastructure company.” What are you in plain English?
We use our data assets and AI to create an intelligence layer that enterprises couldn’t build from the data inside their four walls.
Okay, well, back to the open web, publishers are suffering because of traffic declines due to AI. What does that mean for your publisher-focused businesses, like LiveIntent and Disqus?
Today, because we own a series of assets that help publishers monetize, those businesses have actually been growing meaningfully because everybody’s trying to figure out how to drive incremental traffic and how to monetize it.
That’s today. But what about longer term?
I think the internet splits in two: the HTTP layer, which is visible to all, and an agentic layer underneath where agents negotiate with one another. The HTTP layer won’t go away, because it’s getting cheaper to run. If it costs an airline, say, $10 million a year to manage its site today, it’ll cost $1 million in three years.
Underneath that, we’ll have an agentic layer where agents meet to negotiate. I don’t want to get too out there on this, but we are already starting to see agent socialization. One of the reasons we bought Senso is because that reality is going to exist.
So, 19 deals in 19 years. Should we expect a twentieth?
It’s highly probable. I don’t think we have any holes, though. Senso and Digital Audience closed the last two gaps: identity and a way for our customers to influence AI agents. AI multiples are very high right now, especially in the private market, but as they normalize, there could be opportunities. Nothing on our radar now.
Having done so many deals, any big-picture takeaways to share about ad tech M&A?
I think companies bifurcate. There are those that create intelligence from proprietary data and turn it into ROI, and there are workflow tools with no data and no intelligence layer. AI levels the playing field on public data, so proprietary data is what changes the game.
This interview has been lightly edited and condensed.
For more articles featuring David Steinberg, click here.
