Prepared remarks
Good day, and welcome to the Zeta Global Second Quarter 2026 Earnings Call. Today's conference is being recorded. At this time, I would like to turn the conference over to Trey Campbell. Please go ahead, sir.
Thank you, operator. Hello, everyone, and thank you for joining us for Zeta's Second Quarter 2026 Conference Call. Today's presentation and earnings are available on Zeta's Investor Relations website at investors.zetaglobal.com, where you will also find links to our SEC filings along with other information about Zeta. Joining me on the call today are David A. Steinberg, Zeta's Co-Founder, Chairman and Chief Executive Officer, and Christopher E. Greiner, Zeta's Chief Financial Officer. Before we begin, I would like to remind everyone that statements made on this call, as well as in the presentation and earnings release, contain forward-looking statements regarding our financial outlook, business plans and objectives, and other future events and developments, including statements about the market potential of our products, potential competition, revenues of our products, and our goals and strategies. These statements are subject to risks and uncertainties that may cause actual results to differ materially from those projected. These risks and uncertainties include those described in the company's earnings release and other filings with the SEC and speak only as of today's date. In addition, our discussion today will include references to certain supplemental non-GAAP financial measures, which should be considered in addition to and not as a substitute for our GAAP results. We use these non-GAAP measures in managing our business and believe they provide useful information for our investors. Reconciliations of the non-GAAP measures to the corresponding GAAP measures, where appropriate, can be found in the earnings presentation available on our website as well as our earnings release and other filings with the SEC. With that, I will now turn the call over to David.
Thank you, Trey. Welcome to the team. We are very excited to have you. Good afternoon, everyone, and thank you for joining us today. I will start with the headline. We delivered our 20th consecutive beat-and-raise quarter, delivering the rule of 64 as a company and the rule of 49 excluding M&A. In the second quarter, our year-over-year revenue growth accelerated to 44%, up from 35% in the second quarter of last year. This is the result of Zeta increasingly becoming the system of intelligence for our customers. The second quarter performance once again shows this strategy is working. Second quarter revenue was $443 million, representing year-over-year growth of 44%. That is up 28% year over year excluding M&A revenue. Adjusted EBITDA was $92 million, up 56% year over year with margin expanding 170 basis points year over year to 20.7%. And based upon this strength, we are once again raising the midpoint of our 2026 revenue guidance by $33 million. 20 consecutive beat-and-raise quarters reflects more than just strong execution. It is evidence of growing and durable demand for our platform that turns data into intelligence, intelligence into decisions, and decisions into measurable outcomes. Investors have historically viewed Zeta as a marketing technology company. Marketing is where our platform was first applied and where we built our leadership position. But that description no longer fully encapsulates who we are today. Zeta has evolved into an intelligent AI infrastructure platform. Marketing is our first application, not our limit. Our platform combines proprietary data, AI, workflow automation, and activation into a single operating system that helps enterprises make better real-time decisions and take action. The foundation is our data cloud, built on proprietary data covering more than 535 million individuals globally, trillions of signals, and 20 years of model tuning. And to be clear, this data is owned, not rented. Because nothing you rent can be a moat around your business. Athena serves as the intelligence layer allowing customers to interact with that foundation using natural language and embed AI directly into their everyday workflows. It truly allows customers to focus on outcomes versus navigating the platform. Our activation platform then turns those insights into actions across channels while continuously learning and getting smarter via the results. This has contributed to the strength of our net revenue retention. And with Zeta Business Intelligence, a.k.a. ZBI, we are also extending beyond helping enterprises to acquire, grow, and retain customers to a fourth use case: enabling organizations to transform business and customer data into intelligence, insights, and real-time actions. To be clear, this is not static dashboards or old-school business intelligence used to explain what happened. ZBI helps predict what happens next and acts on it in real time. A major sports and entertainment company uses ZBI today to understand its fans, share of entertainment spending, engagement across live events and streaming. This helps quantify the value of current distribution partnerships, identify the most attractive future streaming relationships, and strengthen its negotiations at renewal. A leading energy drink brand uses ZBI to quantify the incremental value it drives for retail partners, including customer growth, spending, and long-term value. This helps demonstrate the brand's impact to retailers and identify opportunities to deepen distribution and partnership investment. Taken together, these capabilities position us as the intelligent AI infrastructure layer that sits at the center of enterprise decision-making. This transformation has been accelerated by four strategic catalysts. First is our OpenAI partnership. Our collaboration with OpenAI validates the AI strategy we have been building for years, enhances Athena's capabilities, and creates new commercial opportunities. As foundation models continue to improve and become more widely available, we believe the durable advantage will come from what is built around them: proprietary data, governance, workflows, and decisioning systems that turn intelligence into outcomes. Those decisioning systems are our own inference models, smaller purpose-built models that we have been training on the data cloud since 2017. OpenAI powers Athena's voice. The decisions run on our models, and no large language models ever touch the data in our data cloud. The second is Athena. Launched earlier this year, Athena is fundamentally changing how customers interact with the Zeta platform. Athena is true voice enablement and fully conversational. Instead of navigating dashboards or conducting manual analysis, customers can simply ask for outcomes. The platform knows their business, decides, acts, and learns from every single result. That dramatically lowers the barrier to adoption and expands use across teams, channels, and use cases. We have seen that users who interact with Athena through voice exemplify this rapid adoption. Over the last 60 days, Athena's voice users are interacting with the platform at a 500% higher level than non-Athena users. We are also realizing the benefits of AI within Zeta. In the second quarter, 90% of our new code generated was automated, helping our teams to innovate faster and continuously enhance the platform for our customers. The third is Snowflake. We deepened our existing partnership with Snowflake, and we now have over 100 shared customers, reinforcing Zeta's role in the enterprise data ecosystem where customers increasingly want to connect data environments with intelligence and activation. Zeta is uniquely positioned not only to help customers store and analyze data, but to act on it. And the fourth is Palantir. Our partnership with Palantir significantly expands Zeta's enterprise opportunity. Palantir provides the ontology, governance, and enterprise AI infrastructure that large organizations require. Zeta contributes proprietary customer intelligence, proprietary data, identity, decisioning, and activation. And we have reached two important milestones in this partnership already. Our data cloud was fully integrated with Foundry as of July 31, and we have already received multiple agreements for our initial combined sale with several other meaningful opportunities in flight. Together, these four catalysts are accelerating Zeta's transformation and strengthening our position as the intelligent AI infrastructure layer for the enterprise. We believe the defining characteristic of an infrastructure platform is not just innovative technology: it is a platform that becomes more valuable the more customers use it. Every decision it powers makes the next one better. And because those decisions run on our own inference models, not rented tokens, we make millions of them a day with infrastructure economics. That is the flywheel driving our results. And we are seeing it play out across our existing customer base. A leading telecommunications provider recently expanded its relationship with Zeta into real-time personalization, adding grow and retain use cases to its existing footprint. It is a powerful example of our land-and-expand model. Start with one use case, prove value, and expand across the customer life cycle. Athena is beginning to accelerate that dynamic to give investors greater visibility into its impact. Christopher will provide more detail on our initial framework for measuring key metrics like revenue contribution, adoption, and usage later in this call. We will continue to expand on our reporting around these metrics in the back half of this year. GAP is a powerful example of our platform evolution. As part of its broader AI-led transformation, GAP selected Zeta to help architect its next-generation marketing stack with Athena at the center of how customer data, decisions, and execution come together. Our role extends well beyond powering individual campaigns. We are helping GAP to remove silos, unify decision-making across its iconic brands, and build a more intelligent marketing engine that learns, adapts, and acts in real time under a multiyear agreement as GAP's system of record. That is what it means for Zeta to become the transformation agent for the enterprise. This momentum was also evident at Cannes Lions, which served as an important proof point for the growing market interest in Zeta and Athena. We launched Athena for agencies, hosted more than 100 executive meetings, and delivered 37 live Athena demonstrations, resulting in a record sales pipeline coming out of the event. The conversations we began in Cannes will continue at Zeta Live, where we will provide an even deeper look at the future of Zeta and intelligent AI infrastructure. Zeta Live will be in New York City on October 8. I am excited to announce we already have two incredible headline speakers: Olympic gold medalist, entrepreneur, and founder Lindsey Vonn, and entrepreneur and global superstar Kevin Hart. Zeta Live has always been our opportunity to show customers, partners, and investors where we are going next. This year, we expect to introduce the next generation of Athena, continuing the extension of Athena as a superintelligent agent answering questions and empowering workflows as a system of intelligence that knows, decides, acts, and learns. The opportunity in front of us is no longer just about monetizing marketing. It is about helping enterprises turn fragmented data into intelligence, intelligence into decisions, and decisions into measurable growth and cost savings in their businesses. The past year has been an inflection point for Zeta, bringing together capabilities and investments we have been building for years. Yet we are still in the very early stages of this opportunity. AI is changing how enterprises operate, how software is consumed, and what businesses expect from their technology platforms. For Zeta, that creates an opportunity to expand our role from helping customers execute marketing programs to becoming the intelligence layer that enables them to move faster, make smarter decisions, and drive better outcomes across the enterprise. We continue to be the disruptor in this new ecosystem. As always, I want to thank our customers, partners, and shareholders for their continued support, and to Team Zeta, thank you for your hard work, your commitment, and belief in what we are building together. Now let me turn it over to Christopher to discuss our results in greater detail. Christopher?
Thank you, David. I will echo by welcoming Trey to the team. As we announced last week, Trey is perfectly suited to help lead Zeta's evolution into an intelligent AI infrastructure company. And I am thrilled to see Matthew Pfau step up to lead FP&A for Zeta. In my conversations with investors, one thing is clear: a new framework for investing in companies is emerging. Investors are prioritizing their time with companies gaining share and delivering durable, predictable growth. They are increasingly screening for companies generating free cash flow and positive GAAP earnings. And they are ultimately backing companies with defensible AI moats proven by results. Zeta embodies each of these characteristics. And the second quarter's results make this evident. Q2 was our 21st straight quarter of greater than 20% revenue growth excluding M&A and political candidate revenue. Our increased second-half revenue guidance continues that trend. Q2 was also our highest ever free cash flow, paired with positive GAAP net income, showcasing the quality of our earnings and driving the largest full-year guidance raise to free cash flow and GAAP EPS in our history. And we are listening to shareholders. We are introducing an initial framework to measure adoption and monetization of Zeta's AI, one underpinned by revenue. Because moats are ultimately proven by how long customers stay and how much more they spend over time. I will cover all of this in detail along with updates on our pipeline, sales productivity, and drivers behind our increased Q3 and full-year guidance. Let's start first at the top line. For Q2, revenue came in at $443 million, up 44% year over year or 28% excluding M&A. That beat our guidance by $23 million, or 5%, driven by faster growth in both superscaled customer count and ARPU, each exceeding the growth rates in our 2028 model. Superscaled customers grew to 197, up 17% year over year, more than double our 4% to 8% 2028 model growth rate, driven by demand for Athena. Customer gains were especially strong in consumer and retail, telecom, and health care. Superscaled quarterly ARPU expanded to $1.8 million, also up 17% year to year and above our 12% to 16% long-term 2028 model. And we are seeing some interesting usage dynamics unfold in ARPU. First, theme engagement is increasingly voice-first. 83% of customer interactions are now spoken, reinforcing our thesis that natural language will become the primary interface for marketing and business intelligence use cases and has the propensity to drive higher utilization on the platform. Second, the OneZeta sales initiative is gaining speed. Customers using more than one use case are up 90% year over year. Customers using five or more channels are up more than 50% year over year. Cross-sell and upsell deals won in the quarter were up 43%. And we saw double-digit revenue growth across email, CTV, and social, as well as double-digit revenue growth across all three marketing use cases: retain, grow, and acquire. And third, demand for the platform was broad-based across industries. Eight of our top ten grew more than 20% year over year on a trailing 12-month basis, with consumer and retail, financial services, automotive, and health care all accelerating from last quarter. Also notable in the quarter was the significant expansion of the sales pipeline and strength in seller productivity. At the end of Q2, the total sales pipeline was up more than 60% year over year and up over $100 million compared to just 90 days ago. On a per-seller basis, pipeline creation is up more than 100% year over year, as OneZeta and Marigold cross-selling take hold. This is driving higher average contract values on deals won in the quarter, up more than 40% compared to last year. And overall deal sizes in the pipeline increased more than 25% year over year, driven by higher attachment rates across channels and use cases. All while quota-carrying headcount increased by just one versus last quarter to 198, up 11% year over year. That gap between pipeline growth and headcount growth speaks to the strength of sales productivity. And importantly, this growth and revenue upside came with impressive operating leverage. Specifically, we generated $92 million of adjusted EBITDA, up 56% year to year and a margin of 20.7%, an increase of 170 basis points versus last year and $5 million better than the midpoint of our guidance. Marigold restructuring actions and integration savings drove total Zeta expense-to-revenue ratio efficiencies across R&D, G&A, and sales and marketing, improving 318 and 250 basis points year over year, respectively. GAAP cost of revenue came in at 41% as expected. This was 10 basis points better sequentially and 300 basis points higher year over year, reflecting strong social channel adoption by agencies. Cash grew even faster than adjusted EBITDA in Q2. Net cash provided by operating activities was $69 million, up 65% year over year with free cash flow of $58 million, up 73% year over year, a margin of 13.1%, and free cash flow conversion of 63%. We also generated positive GAAP net income in the second quarter of $8.2 million compared to a net loss of $12.8 million in the same quarter last year, resulting in GAAP earnings per share of $0.03. In the second quarter, we prioritized using cash to repurchase shares, deploying $29.9 million to buy back 1.6 million shares. And year-to-date, as of July 30, we have spent $74.6 million on share repurchases with approximately $89.4 million remaining on our authorization. Dilution in the quarter was just 0.1%, and we remain on track to hit our normal course net dilution target of 3% to 4% for 2026. Finally, we closed a new $1 billion credit facility including a $250 million Term Loan A and a $750 million revolving credit facility that remains undrawn, giving us the capital flexibility for M&A, share repurchases, and disciplined investment. To that end, we have been investing in AI for nearly a decade. So today's introduction of metrics showing how AI adoption drives deeper platform usage, longer customer relationships, and higher net revenue retention is not new; it is just more visible than ever. Superscaled customer adoption of AI is ramping nicely. Since Athena's launch for enterprise customers 130 days ago, more than 40% of our superscaled customers are already monthly users, and together they have generated thousands of campaigns using Athena. Customers who comprehensively adopt our AI—using it for audience creation, activation, and other means—contribute a disproportionate share of revenue. This shows up in several ways. Across our total customer base, including 20% of customers who have comprehensively adopted our AI tools account for roughly 70% of revenue. Among superscaled customers, the 50% who have comprehensively adopted our AI tools drive 75% of superscaled customer revenue. And these AI super users grew four times faster than the 80% of customers still early in their AI adoption journey. But it is not only AI adoption that is ramping; it is also leading to longer customer relationships. Superscaled customer relationships now average 56 months, up from 48 months a couple of years ago, and that is based on data going back to 2018. In addition to longer customer relationships, we are also seeing them spend more, generating higher net revenue retention among AI adopters. Customers who have most comprehensively adopted our AI tools have a year-to-date net revenue retention that is 400 basis points above overall Zeta and more than 20 percentage points above customers still ramping in their adoption of our AI, showing that adoption is translating into stronger, more durable expansion. With tailwinds from AI adoption, higher sales productivity, and strong operating leverage, we are once again raising our top- and bottom-line guidance for the third quarter and full year. In doing so, we will maintain our typical conservatism, building in a 2% to 5% cushion that assumes minimal go-get revenue from partnerships and uses 2028 model growth rates for customer and ARPU growth. To be clear, if new customer additions and ARPU growth exceed our 2028 model growth rates, driven by rapid Athena AI adoption or newly announced partnerships, that would push us towards the high end of that 2% to 5% range. For the full year 2026, we are increasing the midpoint of our revenue guidance by $33 million to $1.818 billion, representing growth of 39% or 25% year over year excluding M&A and political candidate revenue. For the third quarter, we now expect revenue of $471 million at the midpoint, up $10 million from our prior guidance, representing growth of 40% or 23% excluding M&A and political candidate revenue. You will note we are continuing to maintain our original second-half guidance for political candidate revenue in Q3 and Q4 of $7 million and $8 million respectively. For adjusted EBITDA, we are increasing the midpoint of our 2026 guidance to $405 million, up $8 million from our prior guidance, representing a year-over-year increase of 45% and a margin of 22.3%, up 90 basis points year to year. For the third quarter, we now expect adjusted EBITDA of $115 million at the midpoint, up $3 million versus our previous guidance. We are also increasing the midpoint of our 2026 free cash flow guidance to $255 million, $20 million higher than our previous full-year guidance, representing year-over-year growth of 55% and a margin of 14% and a conversion of 63%, tracking well towards our 2028 conversion target of 65%. Finally, we are raising our full-year GAAP EPS guidance to a midpoint of $0.10, well above our prior range of $0.02 to $0.04. It is worth noting this guidance increase excludes the impact of a potential one-time tax benefit from the release of a valuation allowance that has a reasonable probability of occurring later this year, which could represent additional material one-time upside. With that, I will close where I began. Demand for Zeta is robust and durable, and we are expanding our platform organically and through partnerships to accelerate share gains. That durable demand and record pipeline improves our visibility, giving us the confidence to raise guidance across the board. That visibility supports disciplined investment, and our balance sheet is well positioned to support our growth priorities. Focused execution drives greater profitability and cash generation, which has us pacing ahead of our long-term 2028 model. With that, I will hand the call back to the operator, so David and I can take your questions.
Questions and answers
If you would like to ask a question, please signal by pressing star 1 on your telephone keypad. If you are using a speakerphone, please make sure your mute function is turned off to allow your signal to reach our equipment. A voice prompt on the phone line will indicate when your line is open. Again, press star 1 to ask a question. And our first question comes from Jason Kreyer with Craig-Hallum.
Great. Thank you, guys. Congratulations. Another great quarter. Christopher, you talked about OneZeta and how that is accelerating the adoption of multiple use cases. I think you said 90% growth in multi-use-case customers in the quarter. What do you think that looks like over the next several quarters? How does Athena continue to optimize customers and maximize both channels and use cases?
Great question. And thanks, Jason. The short answer is I think we've got continued runway because of OneZeta and the Marigold cross-selling continuing to drive both multi-use-case adoption, which as you said was up 90% year over year, but also five or more channel usage was up 50%, and cross-sell and upsell deals within the quarter itself are up 43%. If you look at the trend for ARPU growth over the last couple of quarters, it's been nicely above our 12% to 16% long-term model, and I see that continuing. And we are seeing Athena really just the tip of the spear. It is just getting started. It is one of the single most powerful tools we have ever had for cross-selling across use cases and channels, Jason.
Maybe a follow-up for you, David. On the ZBI, can you just talk about how you go to market with customers? What does that cross-sell conversation look like? And what are the key functions in ZBI that are really speaking to customers?
It is interesting because we have had customers that have been using the ZBI for a few months now before we announced it because we really wanted to get use cases in and try it. So the use cases I gave in my prepared remarks are real-world use cases that they are paying us to develop and create with them. What we are seeing is that customers no longer want static business intelligence. They are not looking to have large numbers of people come in, put data into a user interface, and spit out a static report. What they are looking for is how they make real-time decisions, how they better negotiate on contracts that they are in flux on, like broadcasting for sports networks; how they figure out how to better invest marketing dollars and develop retail partnerships, like with energy drink companies, and so on. So we are actually getting pulled into the ZBI use cases more than we have been selling them. And we are now doing a combination of both together.
Jason and investors on the call, we have included slides in the earnings supplemental to bring those existing customer BI use cases to life. It's actually slides 32 through 38, and David mentioned there are dozens of customers spanning real estate intelligence, customer experience, market sizing, loyalty growth, business expansion, and business measurement, just to name a few.
Wonderful. Thank you, guys.
If you find that your question has been answered, you may remove your cell from the queue by pressing star 2. We will go next to Matt Swanson with RBC Capital Markets.
Great. Thank you so much. David, we talk a lot about that transition from Zeta who to Zeta why to Zeta now. Can you just talk a little bit about when you get names like OpenAI and Palantir and Snowflake—it's probably about as good of a trio of partners in this day and age as you could find—how that helps you in your go-to-market process to skip the evangelizing, but also what it can do for initial deal sizes?
Yeah, thanks, Matthew. I would tell you that it has been game changing for us. The two engagements that we signed with Palantir happened in real time. The Palantir team walked us into two of the largest advertisers in the world, and literally, we walked out with some of the largest test cases and use cases we have ever had as a company based on the fact that they chose us to be their marketing platform partner. If you listen to Alex Karp's remarks, they are over-indexing on partnerships. One of the interesting things I don't think people understand is that the Palantir relationship will be one of the biggest we have ever done because there is not a boardroom today that does not know who they are and is not interested in use cases they can do with them and with us as their marketing partner. It has become an incredible entry point. And if you look at the number of customers they published recently, that's the number of marketing opportunities we effectively have in partnership with them. Add into that the OpenAI relationship, which has been one of the biggest we have ever signed, and the Snowflake evolution, we are seeing more at-bats than ever. If you saw the 17% growth in superscaled customer count, that is not even the beginning of what we're going to be doing with these new partnerships. In the evolution from Zeta Who to Zeta Now, these are game changing even when we are not in the room. When we are out there, we are coming into meetings now and the first thing I get is, 'Wow. How did you get that deal done?' And I have to say to them, 'Which one?' and we all laugh and move on. But I think this is the beginning of a beautiful friendship.
That is great. And then I think we touched on it a little bit when you were talking about the ZBI and the different uses for data. But when you talked about how your platform is currently being utilized for marketing but it is not necessarily a marketing platform, how far do you think some of those adjacencies can go in terms of use cases? Have you seen within any of your larger customers any use cases that really made you think about product development?
Every day we are getting asked questions by our largest clients: 'Can you help us fix this problem or answer this question in a way that we have never thought of?' It is causing an incredible amount of excitement. It is important to note that in Q1 of this year, we were proud to announce that using our internal workflow management tool, Spade, we were able to automate 75% of all new code generated. If you look at the update, the actual number for Q2 was 89.6%; I rounded it to 90% in my prepared remarks. But this is changing the game for how we take interaction with clients and turn it into product development almost in real time. If you look at the sports league example we mentioned earlier, their CMO asked how to better negotiate streaming rights and how to equate viewership and mind share across multiple properties. We were able to put together a solution in hours that they believe will result in millions of dollars of incremental revenue in the renegotiation of those streaming contracts. So it becomes a flywheel because then we get part of the marketing for those streaming rights. It really becomes a major flywheel across the company, Matthew. Thank you.
And we will go to our next question from DJ Hynes with Canaccord Genuity.
Hey, thank you guys. Congrats on a nice quarter. David, maybe we can just follow up on the last point. ZBI and Business Intelligence is a massive category. You talk about getting pulled into this, and there is a lot of different directions you could take it. How do you think about formally productizing some of the use cases? How do you sequence which ones make sense first? High level thoughts on the multiyear playbook would be interesting as you think about building out that fourth use case.
Great question, DJ. What we are seeing is that our data cloud is able to help enterprises make better decisions around certain things very quickly. We are seeing big examples inside of retail: what products should move into retail, how to use co-op dollars and marketing dollars to get better shelf space, and how to create deeper relationships with the end customer. It starts with business intelligence around a retail use case, then we identify meaningful opportunities for them to move inventory geographically, which retailers to do more with or less with, and then we end up being able to market to the end user to drive them into the retailer on behalf of the product, and vice versa. So yes, we are seeing logical productization, much like we have talked about in the past where our platform can help almost any vertical. The more industry expertise our salespeople have across the 15 verticals we operate in, the better they can sell and the higher the sales productivity per rep. We are seeing multiple use cases by vertical that we are building into ZBI, and every new request from a customer becomes productized immediately. That goes back to the ability to do 90% of new code automatically generated. So it is building a very interesting flywheel as we grow ZBI, not just from a revenue perspective—by definition, it is our fastest growing use case right now because it started from a smaller number—but also from product momentum as we add client-requested functionalities day to day.
Very clear. Christopher, maybe a more tactical question. Obviously, we are heading into the U.S. midterm cycle. I am just curious what you are seeing in terms of political advertising demand. How much visibility do you have there today and how does that feed into the guidance?
From a guidance perspective, DJ, we have held to that consistent political candidate guide of $7 million in the third quarter and $8 million in the fourth quarter purposely because we wanted what came through in the numbers to be the strength of the core business as it unfolds in the actuals. The visibility is good; we expect it to be robust. It comes into the pipeline pretty late because these programs can spin up quickly and we execute even more quickly. We get paid ahead, which is nice for free cash flow. But we are optimistic that the number we have in there is conservative. For the full-year guide, it maintains our normal level of conservatism. So despite this being the largest raise of the year of $33 million in revenue, it still maintains our normal 2% to 5% cushion.
I think it is also important to note we raised revenue by $33 million and we raised free cash flow by $20 million. So you are seeing a disproportionate percentage of incremental revenue dropping to the bottom line. You saw a 170 basis point increase in operating margin and a meaningful increase in free cash flow at 73% growth.
Thank you, DJ.
And we will move next to Ron Josey with Citi.
Hi. This is Jake Hallock on for Ron Josey. Congrats on the great quarter, and thanks for taking my questions. My first is on the OpenAI partnership. How deeply is OpenAI integrated into Athena and Zeta's broader platform? Could you help us better understand the capabilities Zeta brings to OpenAI's ad operations? Is there a direct revenue opportunity associated with either side of that partnership?
That's a three-part question, Jake. OpenAI powers the voice component of Athena. No large language models ever see the data in our data cloud or our clients' data; we keep all of that totally and completely safe. Our data cloud is now built on Foundry with Palantir's ontology, so you have an interesting architecture. OpenAI personalizes Athena so Athena can get to know the enterprise user better and anticipate their questions, helping them navigate to outcomes instead of navigating a platform the entire way. As it relates to integration with their ad platform, we are actively serving ads and it is a meaningful revenue opportunity that is scaling very quickly. We have been conservative in our guidance and have not included much from Palantir or OpenAI in forward guidance yet.
We have a follow-up meeting with OpenAI next week with 20 people in a room brainstorming what else we can do together. You consider us one of your most important enterprise partnerships; we look at them the same way. Also, you mentioned 90% of new code generated through AI and that Athena operates largely on in-house inference. Could you touch on how those capabilities are changing product development speed?
It is accelerating product development at a pace I never thought possible, Jake. We increased operating margin by 170 basis points while adding engineers in the quarter; we did not eliminate engineers. The fact that we can add horsepower while 90% of all new code is auto-generated is pushing development cycles that would have taken years into months, and new products inside of ZBI from months to hours in some cases. When we get to Zeta Live on October 8, we'll have a massive unveiling I'm incredibly excited about. This is evolving Zeta as a company in a very meaningful way and it is something that would have taken years that we're now doing in months.
And our next question comes from Elizabeth Porter with Morgan Stanley.
Great. Thank you so much. I want to follow up on ZBI and how that is changing some of the market opportunity. We used to talk about Zeta having 1% of the marketing wallet, with an opportunity to be closer to 10%. As you go beyond marketing to broader business intelligence, how should we think about the wallet share that is up for grabs? How does that change with ZBI? And as a follow-up, how should we think about sales cycles as you expand from a CMO-focused product to broader touch points across the organization?
Elizabeth, welcome back. We have missed you. From a ZBI perspective, we are still looking to get to 7% to 10% of our clients' marketing wallet share; that is not changing. We believe we are on our path to building a $10 billion business with a 30% operating margin, with the vast majority dropping to free cash flow, on the marketing component alone. ZBI is opening new opportunities: we were pulled into this direction by clients and the Palantir partnership massively accelerated it by putting our data cloud on top of Foundry and allowing us to onboard client data with their ontology in ways we could not before. We expect two different sales cycles: existing clients will adopt ZBI quickly, often as a third or fourth use case, and new clients — where we might start with the CIO or CTO — could take a bit longer. Our normal product cycle is anywhere from 90 to 180 days; for new clients this may be on the longer end, but for existing customers adoption timelines are much faster than that.
I wanted a quick follow-up on Athena for agencies. It sounds like you had a lot of high interest in France this summer. When we think about the base of agencies, is there a certain proportion that is a good client base for Athena? Any sub-segments? Do you think most agencies could adopt Athena over time? What are you learning about the onboarding cycle that could drive faster penetration?
To answer your question, I thought it would start at midsized agencies as an entry point. We are actually seeing the larger agencies adopt it faster and at scale. It really changes the game for them from a navigation perspective and the ability to showcase our data cloud and drive substantially higher return on marketing spend. As you know, our Forrester data shows 600% to 700% return on marketing spend typically; clients adopting Athena are seeing substantially higher rates. For example, a combination of a large agency and a very large airline produced a 1,400% return on marketing spend after adopting Athena. I thought it would be small to midsize, but it's starting very large and scaling faster than we expected.
We will move next to Arjun Bhatia with William Blair.
Alright. Thank you, and congrats on a great quarter. David, on Palantir: it sounds like you did the tech migration and you are on Foundry, but also you are closing some deals on the cross-sell. What does that pipeline look like into Palantir's commercial customers? Across different verticals, are there any that get you more excited than others from an opportunity perspective?
I'll be honest: I plan on doing the first 20 deals in partnership with Alex Karp, who has become a very good friend. We have been going out together. If I put the numbers of deals we're working into our pipeline it would skew the pipe up too much at this point — it is that big of an opportunity. I'm tracking an internal pipeline that is very large. This may be the biggest opportunity we've ever had from a partnership perspective. We operate across 15 verticals, and none is a massive concentration, so when I look at their client base, there are far more relevant customers than we could ever start with. We closed two deals at a 100% hit rate. We have other very large ones in flight and see this as a meaningful opportunity that is not yet fully baked into our numbers.
That's great to hear. Christopher, maybe one for you. What's your updated view on capital deployment with the new credit facility? You've mentioned M&A before; any change in the type of acquisitions or assets you're looking at?
Christopher pointed to me, so I'll answer. We will continue to focus on our five pillars of M&A; that's not changing. However, smaller deals we've done in the past would not move the needle at this scale anymore. I believe transformative M&A often transforms both companies for the worse, so we're not looking for one huge transformative deal. We'll continue to do small to midsized deals that add human capital, data sources, and products that our clients want. The new credit facility gives us optionality around buybacks and potential M&A. Our cash position is strong and we're projecting a record free cash flow quarter. Last quarter we deployed a large portion of our free cash flow to buybacks, and we will continue to buy back stock at these prices as we believe it is a strong use of capital.
And we will move next to Carolyn Valenti with Goldman Sachs.
Hey, team. Thank you so much for taking the question. New AI products have generally taken a while to ramp in usage for many software companies. Are there a couple of key things you would point out that you think have made your customers adopt new AI tools faster than what we are seeing in the rest of the market?
Carolyn, congratulations and welcome. One of the biggest differentiators is our ability to demonstrate return on investment quickly. Many AI products require large investments in infrastructure and can take years to show ROI. Our marketing use cases show 600% to 700% return on marketing spend effectively day one. ZBI is delivering real-time business decisioning that drives incremental profits in real time. That immediate and measurable ROI accelerates adoption versus companies selling expensive technology that takes years to pay off.
That makes a lot of sense. And one more: we are hearing more companies talk about headless architectures as a way to disrupt the UI or traditional UI. How do you think about this in the context of Zeta? What makes sense for you?
We have been no-code for many years. The headless conversation isn't something I think we need to deal with on either side. It is not a competitive force for us and not something we'd move into quickly for the front end. Our focus is on how much of our new code we can generate and make generally available to clients. We've gone from 75% to about 90% of new code being automatically generated, which is a massive competitive advantage. As competitors try to catch up, we are already moving ahead where they will not be able to catch up for many years. Our no-code architecture has enabled that.
And we will take our next from Terry Tillman with Truist Securities.
Hi, this is Luca Gudes on for Terry. Thanks for taking my questions. Considering the big strategic win with GAP, how is RFP activity and what are you seeing in terms of large MarTech replacement cycles potentially aiding revenue and business in the second half or helping enhance visibility into 2027?
RF activity is very strong. The pipeline stats we shared reflect greater than 60% year-over-year pipeline growth and higher average contract values because more of it is RFPs. I'll turn it to David to speak broadly about what is driving that environment.
We are in the middle of what looks to be a marketing cloud replacement cycle, and GAP is a perfect example. GAP had been with one vendor for quite some time. There were three other vendors we displaced in addition to that vendor to become the system of record at GAP. Large enterprises do not want to use four or five different vendors; they want next-generation technology. Today we are the only marketing cloud that has data and AI as native, foundational elements to the application layer. Other platforms must use APIs and integrations to access AI and data, and that latency destroys return on marketing spend. Our foundational platform can answer in a millisecond what other platforms cannot answer. We are seeing RFP velocity increase and closings accelerate, as we saw this quarter; we have a record pipeline.
If I could sneak one more in: given the increased adoption with Athena and AI-enabled workflows, can you help us think about the puts and takes on gross margin performance in the back half of 2026?
Gross margin performance in the back half of the year is largely dependent upon mix. We talked about how the quarter came in where expected given the direct mix of 72% and how that efficiency plays across R&D, G&A, and S&M. We saw strong adjusted EBITDA in the quarter and expect efficiencies to continue. CapEx efficiencies flowed to higher free cash flow and the actions on dilution and stock-based compensation generated positive GAAP net income.
And we will move next to Jackson Nichols with KeyBanc Capital Markets.
Hey, guys. Thank you for taking the question. David, how are you thinking about the opportunity for Athena-heavy users today post-Foundry infrastructure pivot? What does that mean for heavy users on the new platform and what is the biggest risk in the coming months during the transition to Foundry?
The transition is done and is seamless to clients. We rearchitected the data cloud on top of Foundry and adopted Palantir's ontology, which is a massive benefit. Athena at the top of our tech stack, powered by OpenAI, will interact with the data cloud inside Palantir's architecture, which moves extremely fast. We will be able to answer more questions smarter and faster and onboard client data faster with better orientation, leading to higher levels of intelligence sooner. Traditionally the longer a client works with us the smarter the platform gets; with Athena and Foundry we are seeing that occur faster. Heavy users already show stronger metrics—AI adopters have higher net retention—and I think you'll see that continue to evolve.
That makes sense. Christopher, what kind of usage trends of Athena are baked into the superscaled ARPU growth to achieve the organic guide we have line of sight to?
What we built into guidance is what we have already signed. We are not leaning on go-get assumptions for AI adoption or newly signed partner agreements that are in the pipeline but not yet closed.
And up next is Koji Ikeda with Bank of America.
Great. Thanks. Appreciate you taking the question. Can you put a finer point on what is working well in terms of driving improvements in sales rep productivity? Also, can you help us think about the outlook for quota-carrying rep headcount for the remainder of the year to address the pipeline you talked about?
OneZeta continues to gain speed; we launched it roughly 15 months ago and we're starting to see the benefit where sellers attach more channels and more use cases into deals. Pipeline creation per rep is up around 100%, which is a big source of productivity. Six months after acquiring Marigold, the hard integration work is done, and we are starting to see benefits of cross-sell and upsell activity, especially around loyalty products and selling Zeta's acquire and grow use cases. We have the right tenure mix of reps—those in their first 12 months, 12 to 24 months, and greater than 24 months—so the productivity improvements we see can continue through the year.
I also think our hyperfocusing by vertical and hiring salespeople with industry expertise has unlocked a major opportunity. When we bring in people with industry expertise, productivity goes through the roof.
And we will go next to Richard Baldry with ROTH Capital Partners.
Thanks. On the generative engine optimization (GEO) side, do you think that is helping you win client wallet share, or is that dollar moving from one channel to another? Also, you have grown without adding a lot of headcount in recent years. Do you think there is a point when to keep up with these opportunities you need to add more headcount?
We're seeing GEO as a real upside. We are one of the few companies with API integrations into Claude, ChatGPT, and Gemini, so we can serve across all of those in real time through one GEO user interface and serve marketing into Gemini and OpenAI. New customers see that as a 'shiny new thing' and existing customers adopt it as part of their strategy. Regarding headcount, we are not seeing the need to add a lot more right now. Product velocity is incredibly high: 90% of new code is auto-generated. We signed an enterprise agreement with OpenAI giving all employees access to their products with managed usage and approvals. Our internal AI unit costs are well under 1% of revenue and are decreasing. We will add headcount, but at a slower pace than revenue growth, continuing to grow EBITDA and free cash flow faster than revenue.
We have actually had headcount grow while revenue has increased much faster, and even though AI-based usage is up, unit costs for AI have declined almost 40% due to negotiation and scale.
And we will move next to Scott Randolph Berg with Needham.
Hi, everyone. Nice quarter. My question is on your expansion opportunity within your superscaled customers. The slide in your deck showed ARPU has been range-bound over the last seven quarters between $1.6 million to $1.8 million. We know you're expanding with some customers. I'm trying to better understand that dynamic. Are new customers coming in at slightly lower ARPU, balancing out expansions? Are some expected expansions in the back half not yet reflected?
It is more the latter. Slide 8 in our earnings supplemental helps explain this. The metric is masked because as new pilots and proofs-of-concept become superscaled (cross the $1 million threshold), they are still a distance from the ARPU of our more mature superscaled customers who have been on platform longer. The average ARPU for customers on platform less than 12 months is about $700k, while customers on platform more than four years approach $4 million. So the mix of newly superscaled customers and long-tenured customers explains the range and why superscaled ARPU can appear range-bound even while expansions are occurring behind the scenes.
Thanks for taking the question. I'll jump back in the queue.
And we will go next to Clark Wright with D.A. Davidson.
Great. Thank you. It was great to see better-than-expected organic growth results this quarter. How much of this growth is coming from continued success with agencies versus your direct enterprise sales motion?
It is well spread out. The agency business remains around 20% of revenue and direct enterprise about 80%. We have not seen a meaningful skew change. We are seeing meaningful organic growth across both components.
There was a sequential step down in direct revenue from 75% to 72%. What caused that this quarter and should we expect that level going forward?
That was driven by new sales and expansions with agencies within the quarter. Those newer agency signings tend to begin with social, and social grew very rapidly in the quarter, which drove the integrated platform revenue mix higher and direct-platform mix to 72%. That mix was expected given the pipeline and is driven by very strong agency adoption of social as the initial channel.
And we will go next to Naved Khan with B. Riley Securities. Your line is open.
Hi there. This is Ethan Widell calling in for Naved. Thanks for taking my questions. To start, it's great to see strong uptake with Athena since making it generally available. It sounds like new costs could scale with usage. If usage continues to scale, how should we think about cost scaling or token costs? Do you pass that directly onto customers or does usage become a cost of revenue consideration?
We have an agreement with OpenAI that gives us tremendous visibility into cost. This product is focused on a license for the product embedded into Athena rather than direct token utilization. I feel comfortable we will keep total AI cost well under 1% of revenue while growing revenue at substantially faster paces.
Separately, eight of your top ten industries grew over 20%. Can you speak to how those areas are pacing so far in June and July?
Performance in Q2 was pretty linear and I'd expect Q3 to be similar. I won't project a single month, but the momentum is clear. Eight of the top ten grew over 20% and one of the two that did not was advocacy, which has a lot of tailwinds going into the second half. Several industries accelerated from a trailing 12-month basis ending in June: consumer and retail, financial services, automotive, and health care.
We would not have raised the quarter by $10 million if we did not think we had visibility into the quarter. Nothing in July led us to believe we shouldn't raise the quarter and the year as much as we did.
Understood. Thanks for the color and congrats on the strong results.
That concludes our Q&A session today.
I will turn the conference back to David A. Steinberg for closing remarks. I just wanted to close on how incredibly proud I am of the Zeta team. To be able to continue to execute: 20 quarters as a public company, 20 quarters beating and raising. To continue to execute over that period of time with that level of excellence, to continue to see accelerated sales growth, and to be able to do partnerships with three of the world's most important companies within just a few months. If you had told me a few years ago that we would be in a position to announce partnerships like OpenAI, Palantir, and Snowflake, I would have been blown away. That shows what has happened to Zeta as a brand because none of them would have trusted us if they did not trust our brand and our business. So thank you again to all of our Zeta people, to all of our clients, and especially to our partners. We appreciate everything that you are doing for us and with us as a company. Have a nice day, everybody.
That concludes today's call. Thank you for your participation. You may now disconnect.