All right. Thank you everyone for joining us. We've got Viant Technology here. Nick, thank you so much for joining me today.
Yeah, happy to be here.
Maybe the first thing we could get off, like AI is clearly reshaping the industry. You guys have leaned into this more than, you know, most peers obviously in the industry with your ViantAI platform.
Yeah.
and building really like an autonomous advertising solution. Could you maybe take a step back and just talk about the products that you've released in ViantAI and just what each of them brings to the table and what you're trying to do?
Yeah. I'll take you guys on a little bit of a journey here. Our long track record has been to go after autonomous advertising. It started about a little bit over two years ago. It started with a product called AI Bidding, effectively that's algorithmic purchasing. We are taking over what the trader does. We're bidding on their behalf. An oversimplification of kind of how that works is if you had a trader that was building a plan and they wanted to buy Hulu ad inventory, for example, they would obviously type in their budget and then specific parameters, like, "I want to target a certain audience, I want to target a certain content, and then I want to cap my CPM or my bid is maybe like $30.
That's my bid. If you opt into AI Bidding, then you give Viant some control to go ahead and do bidding on your behalf, right? Basically, then you're allowing Viant to constantly change the bid, prod the SSPs looking for the floor price to try to find anything lower than the $25 bid you submitted, but still satisfies all the parameters that you originally put in. When we do that, when we can achieve that, we'll get a spread, right? If the advertiser wanted a max bid of 25 and we were able to get the bid done at 15 whilst again still fulfilling the parameters, we get to keep some of that spread. It's a win-win for both parties. The advertiser gets better return on ad spend, and we get a higher take rate.
That product is, like I said, just over two years old. 85% of the spend on our platform now utilizes AI Bidding. From there, in September 2024, we announced AI Planning. Basically, this is super relevant right now because everybody's talking about cutting the heads off of these software business models, right? Effectively, we did this starting in September 2024 with AI Planning. We said the DSP interface is complex. It's too hard if we want to go after the 10 million advertisers that Facebook has, that Meta has, right? If you were just to take a step back, if you were to enter into a DSP right now, if you were to open it looks like a Bloomberg terminal effectively. Now, you guys would all be familiar with that because you operate a Bloomberg terminal.
If I put you in front of a DSP, you would have no idea what you're looking at, just like a non-finance person when they would go into a Bloomberg terminal, they would have no idea what they're looking at. We kind of chopped off that layer. It still exists, but with AI Planning, we chopped it off such that if you just want to type in a very prompt-like format, "I'm Walgreens. I've got $50 million in ad spend. I want to deploy in the fourth quarter, and I want to drive foot traffic," instantly, within 30 seconds, a plan is built for you. That plan knows, oh, you're Walgreens. I know who your competitors are. I know where your stores are. I know who your target demographic is.
It's going to build the plan and allocate $ to CTV and certain publishers and do frequency capping and select which CPMs should be allocated to what specific channel. The whole plan is built for you, and if you don't like it or if you don't have a digital creative, right? It's built a digital creative component for you just tell it. You just simply type it in that format. "I don't have a digital out-of-home. Rebuild." Boom, the entire plan is instantly made for you once again. It really in a way alleviates what currently goes on with media planners. Having to hire, you know, media planners to build a plan over a two to three week process, we can do that instantly now leveraging AI.
The next product was analysis and measurement. That was June 2025. Similar to what I just described, now instead of getting a 50-page report, or hiring data scientists to sift through that to understand how your campaign's performed, you just ask. You just ask the question, just like I described previously. What's my top-performing campaign? How's my campaign performing in the Midwest? You get instant answers. Finally, what just rolled out January 1, in conjunction with CES, is AI Decisioning. That is an amalgamation of everything I just described, right?
Except with the added benefit of dynamic delivery, meaning if you built the plan and you agreed to it and you submitted it, and as that campaign is now going, AI will read the signals that are out there in the market to change the plan to deliver the best outcome. If you started the campaign and CTV CPMs go down, right, and the AI reads that via the signals, it can adjust the campaign to increase your CTV mix to deliver a better outcome. That, of course, was announced in conjunction with our Outcomes product, which goes after performance advertisers, which I'm assuming we'll get to. Yeah. The whole AI suite is now out. It's been a four-phase plan, and effectively it's been put together to deliver autonomous advertising, and that is now live at Viant beginning January one.
I think the big question that everyone's asking across the entirety of ad tech is why Viant or why any company? You know, in a world where software and UI are no longer moats.
Yeah.
What are the structural assets that Viant has that differentiate it from any other DSP in the market?
Just to be clear, like we were trying to tear down that moat that did exist. Like right now, you know, we're trying to win spend from The Trade Desk or a Yahoo or a Google DV360, right? The difficult part about pulling spend from that, from those DSPs has been that they're complex, just like our DSP is complex to use or historically was. We need to basically take people that had learned the user interface and pull them onto ours. Like that software that you're talking about was a hindrance in getting them over. Obviously when we rolled out AI Planning and Outcomes, that barrier comes way down as you're kind of seeing the threat that exists out there right now. Behind that interface, right, is a bunch of proprietary data and high fidelity unique signals.
That has always been the true asset of Viant, right? It's just been hard for everybody to utilize because they had to learn our DSP in order to get familiar with it. What are these proprietary signals? It all starts with Household ID, right? That is the industry leading identifier in the marketplace across the open internet. If a bid were to come in from a SSP or a publisher, 80% of the time in total and 90% of the time on CTV, I can read that bid and tie it to a household. I have a household graph that maps 95% of all households, right?
The content that fills that ID graph is data and behavioral characteristics from 70 data partners that we've partnered with that provides us anything from, you know, Experian, TransUnion data, auto intender data. Like that household is characterized by a bunch of data that describes its behavioral characteristics. Basically that signal, that high fidelity unique signal allows me to target audiences better than any other DSP in the marketplace, right? I know this to an extent because if you look at the competitors that are out there, let's take The Trade Desk, right? They use Unified ID 2.0. That's their identification solution. That's an open source product, and I'm a DSP, so I read all the bids that are coming in, and that's available on about 20% of all the bids. LiveRamp is another open source product.
I'm a DSP, I can see when it's coming in. 35% penetration, 30%-35% for LiveRamp. We're at 80% on the open internet and 90% in CTV. If you're an advertiser that cares about targeting a specific audience, I'm saying that we can do it better than any of the competitors that I've just listed. On top of that. That's a unique identifier from an audience standpoint. We also have a unique identifier from a content standpoint, and that's IRIS_ID. Every other DSP, if you were to buy Hulu ad inventory, Disney+ ad inventory, Netflix ad inventory, whatever it is, you buy the app, right? You don't know what specific content you bought against it if you were to buy programmatically. We've partnered with the publishers and layered in IRIS.
They effectively feed us the video file. We get the metadata or scan for metadata against it, pass it along the bid stream such that our advertisers now can target specific content within CTV. For an example, if Cabela's, right, wanted to target effectively Yellowstone season 1, episode 1, we pass that metadata along in the bid stream such that they can target it, and then they could align a fishing commercial or a fishing ad to a fishing scene in Yellowstone season 1, episode 1, right? That's contextually relevant. Then they'll get the data on the back end that shows that that works, right? Then they'll effectively be able to price all inventory accordingly, such that they'll see that there's high attention, high relevance on that Yellowstone content. Maybe a romantic comedy doesn't perform as well. They'll make the decision.
I can either buy Cabela's, and I can not buy the romantic comedy, or I can cascade the pricing accordingly. When I buy Yellowstone, I'll buy it for a $40, $50, $60 CPM 'cause it works so well. Maybe I'll buy that romantic comedy, but I'm not paying any more than a $10 CPM for it based on the data that's given back to us. Unique audience identifier, unique content identifier. We also have supply scoring models, just because we have direct integrations with 75% of all the CTV inventory that's out there going directly to Paramount, directly to Disney. We have that high fidelity signal that allows us to say, "Hey, some of this inventory is garbage," right?
Some of this inventory is fraud, some of this inventory is bots, and we're not going to let our advertisers buy that. That's all unique modeling that we've done based on, you know, a litany of historical campaign data. You take all that together and the advantage is the same. It's that proprietary unique data, that high fidelity signal. Even if you cut the head off the user interface, what we can say if we're competing with a bunch of other AI-driven platforms, is that my AI is better than your AI because it's built on higher fidelity proprietary data.
In this new AI world, it does feel like there's this shift that's taking place between brand and performance. Like, there really actually isn't anything that's brand as much anymore. Everything's becoming performance. Do you think that's a fair characterization? How does Viant fit into that? Obviously, you have ViantAI and everything else, and this is the direction you're going, but is that a fair characterization of the general market?
I think it's a, it's a fair characterization for everybody that is not, like, the top 500 advertisers in the world, the biggest brands. Because I still think, like, a Nike, a P&G, a Unilever, it is, like, when everybody's your customer, like McDonald's, when everybody's your customer and you are everywhere, you just reach in frequency, pray and spray, right? As you go down below the top 500, I think everybody's customer then is a little bit more niche, right? Their target audience. That's exactly where Viant fits in. We don't win McDonald's, but we win Whataburger and Sonic Drive-In. We don't win Nike, but we win New Balance. The reason is because for those brands, their customer is not everyone, and they're not everywhere.
Sonic Drive-In, Whataburger, they don't want to blanket the US in ads because their stores are not all over the US, they want. They utilize Viant and the audience identifiers to target certain audience and target certain geos, right? A great example that we have is, like, a major college or university in Arizona, right? We do really well with universities. A university would never blanket the US in ads, right? That'd be a total waste of ad spend. What they do on Viant, utilizing our platform, is that that university targets a five-state radius around Arizona, looks for $100,000 incomes, looks for 18- to 24-year-old households, but then they deliver a branding campaign. What do you call that? Like, is it performance? I call it sophisticated brand advertising, right?
They still then want to know, okay, you delivered the ad to the right people? Yes. Did they go to the website? Did they look? Yes. Like, that's a mix of brand or performance. I do think though that over time there's going to be more of the top 500, and we'll touch on it in a second, that can find a way of being more sophisticated in the way they go about building ad campaigns, and some of them will start to do that over time.
Let's hit on that, because you landed last quarter Molson Coors, which is a big win for you guys.
Right.
You've talked about this $250 million pipeline of enterprise spend.
Right.
Can you just talk about why are these advertisers coming to Viant?
Just obviously if these if you land a couple of these, it could be really meaningful to the business.
Yeah, I mean, Molson Coors is huge. It's very meaningful for the business, that kicked off in the first quarter of this year. We just won Molson Coors, announced that the tail end of last year. Basically, yeah, you got an advertiser that, like Molson Coors, who's always deployed reach and frequency type advertising, but in this example, I think they too see the need to be a little bit more sophisticated. They manage a portfolio of beverages, right? They're tapping Viant, not only because they want to not exit linear, but they recognize that linear is dying and it's all that audience is moving over to CTV, and when they go to CTV, they have an ability to deploy more sophisticated campaigns.
They have a portfolio of brands, and they want to align each one of those brands to a certain demographic. They're tapping Viant to find that demographic. They have this brand, they want to tie to this demographic. Viant, where are they? Where can I find them when they're available, right? That's where the audience ID, that household identifier really shows its value. Effectively, that, you know, winning Molson Coors was a bake-off. You know, Molson Coors typically, you know, from our perspective or from Wall Street's perspective, they're thinking, oh, they're going to go with DV360 or The Trade Desk, one of the big behemoths.
No, that came down to their need to find a certain audience, to grow penetration within a certain audience and they pitted Household ID against the other identifiers that are out there, and we showed them obviously IRIS_ID as well. In that bake-off, we were just able to find an audience much better than the competitors were, and that's the way they want to deploy targeting here and now. Will that be everybody? Hard to say. Like when P&G, they're selling soap, toothpaste, toilet paper. Do you target anyone with that? Maybe it's just everybody. Like, I think some will always stay big brand advertisers, but there is a cohort, I think, of the big brands that want to deploy more sophisticated campaigns knowing that the technology's out there.
Makes complete sense. Going the opposite direction, I think that ViantAI also opens up that, you know, SMB performance type.
T hat's historically only spent on search and social.
Just where are we today? Does that, does that open up that opportunity, one? Where are we today, and when can that actually start to impact results?
Yeah. That was the whole point of the Outcomes launch. If you think about it, if you take a step back, what I've been saying is that this DSP interface has been too complicated to operate for anybody that's not sophisticated enough to use it. Big brands, big agencies, they can utilize the CTV. If I want to go lower, it's just too complex for the DTCs and the SMBs that are out there, right? They're out, they're on search and social because those user interfaces are just so easy to deploy. You and I could start advertising there instantly. I've always had the targeting capabilities that they need. I've got Household ID, I've got IRIS_ID, I've got supply scoring models. That's all existed. I needed to lower the barrier to entry by eliminating that software layer, right?
Now when a SMB can just type in, you know, "I'm Fuller Sandwich Shop," my, a restaurant I go to a lot downtown, you know, "I've got $20,000, $30,000 deploy in the fourth quarter to drive foot traffic," AI can utilize our signals to carve out a five-mile radius around that restaurant and start advertising to the local community to bring in that traffic. Outcomes was just launched January 1st, and so we're going to go after performance advertisers, effectively the 10 million advertisers that sit on search and social. I think, you know, as far as like when does this start bearing fruit, you know, I would say over time we want to go after the 10 million, but right now we've got about or almost 2,000 advertisers on our platform.
They're allocating spend based on that sophisticated branding label I gave. They all have performance dollars.
Yeah.
They're just going to search and social. Now can I go to them and say, "Hey, pull some of your performance budgets into CTV because CTV is now a performance channel. I have the tools to enable you to do that." What I'm saying is the low-hanging fruit right now is just going after our current customer cohort and have them increase the spend on our platform by bringing in performance dollars.
No, that makes complete sense. There's a debate in the industry and between SSPs and DSPs. This has always been the age-old debate. You start to see it more and more. I mean, you guys have an SPO offering going direct to publishers. You see it the opposite way, where PubMatic and Magnite, they're building ClearLine, they're building Activate, and they're building buyer tools moving the opposite direction. Just could you just talk about like how would you respond to that and just why should the DSP still be around? What are the structural advantages of a DSP?
Yeah. I think overall this is like so simple. An advertiser wants to buy at the lowest price. A publisher wants to sell at the highest price. That will always exist. They've just employed SSPs and DSPs to do the bidding for them and manage the bid stream. Like, the same dynamic still exists, right? Like I can ClearLine to, like, this analogy. It's as if I want to buy a house, right? I hire a realtor to help me find a house, and that realtor tries to sell me the house that she owns. Like, is she going to bite and scratch to get me, like, the lowest price? No way. She owns the house. It's to her benefit to sell it for the highest price.
You know, ClearLine also has always been. Like, the thought on ClearLine has been let's get dollars that are stuck in linear into CTV. These are price-sensitive dollars. These are dollars that don't want to employ targeting, you know, the sophisticated audiences content that we enable. It's pulling in some dollars that are stuck in linear into CTV. What I'm saying is that it's not competing with me at all. I think as those dollars eventually want to get smarter, then they would leave ClearLine and go to a DSP. We are not losing a dime to dollars that are going through ClearLine or Activate. Again, remember, we have the audience data, we have the contextual data, we have the supply scoring data. All that sits on our side.
Really, when you refer to direct access, you know, we don't monetize it is what I'm saying. Like, ClearLine obviously is monetizing some of the buy side. Because in CTV, there's a concentrated number of platforms that matter, we can go to direct because of our size. Like, the advertiser on our platform, they decide.
Yeah.
Why do you want to buy through the middleman? Do you want to buy direct? There's no cost differential other than bypassing some of the SSP take rates.
Yeah.
We're not like OpenPath at The Trade Desk, where they, like, they charge for it.
No, totally. Obviously the big debate in the industry too is Amazon.
They've pushed harder with their DSP. Have you seen that at all? Have you seen the competitive intensity of the industry change at all? Just can you talk to that? 'Cause it's clearly-
Yeah
the clear-
Yeah
Debate point in the industry.
Well, I'm not going to, like, indicate that they're not a threat. Obviously. It's Amazon. They're huge. They've got Prime Video, which is unique content to them. They've got, you know, effectively infinite capital, and they apply some bundling mechanisms that benefit them as well. Obviously a threat, and it's probably, like, the biggest threat that I could possibly think of just in general. There are just a few things to mention there. The first off, Amazon, like DV360, like Yahoo, they are a seller of ads, right? They own Prime Video, and they want to send everything they can to Prime Video, and the reasoning is so obvious. When something goes to Prime Video, they get 100% margin. When that DSP sends something to Disney+, they get 20%, right?
Naturally, they're going to attribute all of the sales, all the performance to Prime Video, right? Because they're incentivized to do that. What we'll show on attribution reports is, "Oh, man, Amazon Prime Video is driving all your sales, not so much Disney." I'd start shifting more budget to Prime Video and out of Disney and Hulu. They'll never be an independent and objective player in that regard, never truly working out, working for the advertiser. There's a few other misconceptions that are out there that are worth clarifying. They talk about a 1% take rate, and if you look at our take rate, and Trade Desk's take rate is north of 20% now, ours is north of Trade Desk.
People are like, "Oh, man, how do you compete with a 1% take rate when you're at 20?" Well, that's not apples to apples. That is very simple advertising. That is a relationship that was consummated outside the DSP and then is just going to go through the DSP from a workflow standpoint. We have programmatic guaranteed deal. We do that as well. It represents 2% or 3% of our ad spend, and we get a 3% take rate relative to their one. It's hardly material, hardly relevant. It's just kind of there. Even when Amazon does price, you got to remember, like, they're trying to sell things to or send budget to Prime Video. What's the take rate when they send something to the Prime Video? Let's say it's zero.
they charge a $70 CPM. What's the difference between a $70 CPM and a $50 CPM with a 20% DSP take rate, right? It's just all the dollars are flowing to them on their own or in entirety anyway. Finally, just the other concept worth highlighting is that everyone talks about, of course, well, Amazon's got to have unbelievable data, right? Unbelievable data. That I'll concede that is for sure the case when you have a Prime Video member or a Prime member watching Prime Video, right? Hook, line, and sinker, they see everything, and they can deliver the best ads possible.
When that same user then leaves Prime Video and is on Paramount+ or Hulu or HBO Max, what signal is available to Amazon that allows them to attach it to their identity data, right? That signal doesn't exist. They don't have a household ID replacement, right? Or a household ID competitor. Now, they could utilize open source products like Unified ID 2.0 or RampID to make that attachment, but that's still, like I said, the mix there is 20%-35%. Household ID at 80%-90% crushes that all day. Anybody who wants to deploy a sophisticated ad campaign outside of Prime Video would still be more inclined to utilize our solution as opposed to Amazon's.
No, it makes complete sense. There's a lot of catalysts for the top line, the $250 million in enterprise spend. The buying AI is rolling out. As you think about costs, what are the incremental investments that you need to make in the business? Should we expect the incremental margins on these increased dollars flowing through at very high incremental margins? Just how should investors think about the cost side of the equation?
High incremental margins, yes. If you think about it, AI costs have been embedded in our structure for three, four plus years now, because when we IPO'd, the whole thought process was to be an advertising, autonomous advertising solution. All of those costs, those building mechanisms, have been built into our structure over time. Last year, or I should say through third quarter, we were targeting around 14% OPEX growth. Embedded in that was the OPEX from two acquisitions. If you strip that out, it's more like 8 % of growth. I don't foresee, like, anything material that would suggest, like, any massive change in that type of outlook on this point, right?
Like, there's no incremental massive AI cost that we haven't already been incurring because we've been going through this AI exercise for four years now. This isn't new to us.
Nick, thank you so much for coming. I appreciate it.
Happy to do it, man. Good stuff. Thank you.
Thank you. Appreciate it.