Jeromy Sonne teaches us about How DSPs Evolve and AI in AdTech

Tim Rowe:
Welcome back to OOH Insider, where you learn about ad tech, DOOH, the larger media landscape, where it all collides. Today, we'll be talking to Jeromy Sonne. He's the founder of daypart.ai. Look back, check out last week's episode with Jonathan Moffie. He's the founder of Streamr.ai, a platform that unlocks CTV long tail demand by enabling publishers to create self-serve ad platforms. Looking ahead, we'll be talking to Bill Harvey next week. He's a research pioneer. He's been in media research for more than six decades. His team of engineers analyzed a million words in the unabridged Oxford Dictionary. They distilled them down to 265 words that Bill calls driver tags. Those driver tags describe behaviors and emotions that empirically lead to choice. This was the tech that drove the early discovery engines on TV, recommendation engines, things like, you watch this, you might like this. And today it helps match the emotional content of an ad to the surrounding context of content it's being consumed in. So we'll be talking to Bill next week. That's going to be a lot of fun. And there are only five weeks left. Placer AI Discover in New York City, Chelsea Pier, December 10th. It's a leadership conference for anyone looking to incorporate location analytics into their solution. It's an all-day event at the iconic Chelsea Piers, Pier 60. You can get 70% off of registration. Go to theOOHinsider.com. You'll see the promo code across the top announcement bar there. You can't miss it. A full price ticket's only $399, but With code OOHInsider70, you'll save 70%, which makes it like 120 bucks. So go to theOOHinsider.com. It'll be linked at the top. It'll be easy to register. I'll see you there. Jeromy, welcome to the show.

Jeromy Sonne: Thank you. I appreciate you having me. I'm always startled that people want to hear anything I have to say, but here I am.

Tim Rowe: They follow you on Twitter, they're connected with you on LinkedIn, but they might not have heard from you like this before. So I'm excited to help bring some of that conversation to the table. Thanks for being willing.

Jeromy Sonne: I like to have a lot of fun on social media. I think people take themselves way too seriously, but I think it resonates too, because everybody else is tired of everyone taking each other way too seriously. I don't know. I've been told there's a lot of helpful information in the jokes when I start getting down into it. So hopefully. Hopefully we can have some fun and also help some folks out. Teach, learn, all that good stuff today.

Tim Rowe: Let's do some learning. Let's learn about how Daypart, your platform, you've gone through a bit of an evolution. Started out as a DSP, evolved into an AI media buying platform, and now you've really zeroed in a focus on B2B, ABM. I would love to talk through everything, strategy, creative, buying, measurement, but maybe start us at, at day zero. Who is Daypart?

Jeromy Sonne: So let's see how I got into advertising was, this is not a joke. completely serious. It's a little embarrassing, but I like to tell the story anyways. So in 2011, I was in college and I was doing ROTC and I was getting ready to go to law school. I was going to be a JAG attorney and all that because that seemed respectable and that was something my parents wanted me to do and all of that. So very, very different path that I'm on now.

Tim Rowe: And I just didn't really want to do it.

Jeromy Sonne: You know, I had a roommate that got deployed and that kind of was like, give me some like strong feelings. And this is nothing against military service or the folks that are in there, you know, great people that are sacrificing a lot, you know, nothing against that just wasn't for me. So I started looking around figuring what should I do instead? And I saw that movie, the social network. And I was like, Oh yeah, I should be, I should be a tech billionaire. That sounds just swell. I'll do it and it'll be great. And it'll be lots of fun. I'll wear a hoodie, I'll do a rip shots of tequila while I like write code and that'll be super fun. So I came up with a startup idea that was basically like Kickstarter, but worse. And I pitched it to a couple of my buddies that were comp sci majors. And we went and tried to do it. And I did a bunch of business plan competitions and we scraped together, I don't know, 20 K or something like that to try and launch this thing and various pitch competitions. And the funny thing is that I, at some point along the way, somebody told me being the idea guy was not a good You should not be the idea guy. You should actually do something in the startup. Just hanging around, having cool startup vibes was not a value add to the business. So I was like, all right. Marketing, that seems like something that I could be into. So I started running Facebook ads in 2011, which was a very, very good time to start running Facebook ads. And so yeah, yeah, the only thing that worked in that business was basically that we were able to acquire a lot of users because I started just figuring out advertising right across that and across other platforms. And I was doing all sorts of guerrilla stuff on Reddit back in the day and things like that, trying to get folks user acquisition, I got really good at it. And then that, when that failed, because it was a terrible, terrible idea. Um, I moved on to, to my buddies and I was like, look, I'm trying to figure out what's next. I don't know what I'm going to do. I was applying for jobs. They were like, well, come consult you. You ran Facebook ads. It did pretty well for me. Right. One of those early companies, uh, was a company based out of, uh, Iowa, uh, in a small town. And it was a education app and there we helped take it to number one grossing in the iTunes place. Wow. Uh, for education using, using primarily Facebook ads and Twitter ads at the time. And then I ran a small agency for a while. It wasn't ever huge, but it was like performance media buying agency. We did pretty well. A lot of mobile app marketing started doing more and more e-commerce as that took over. I was briefly director of marketing for a B2B fintech company based out of Iceland. Uh, which was a cool adventure for sure. Uh, eventually kind of started expanding my horizons out into more programmatic and like additional media. As I was like, these conversion campaigns on meta are going to automate me out of a job. Like your media buying is not going to be as much of a thing with Facebook.

Tim Rowe: When were you thinking about this?

Jeromy Sonne: Like 2016, 2017, when that started and I was like, yeah, I'm going to get automated away. I need to move forward. So I. Started getting into more programmatic buying, started getting more into traditional buying, taking the performance marketing frameworks and ways that I would do things and transmitting it to those worlds. And then decided to cross over to the dark side and start an ad tech company. That was called Accessible. It was a audio ads DSP.

Tim Rowe: Okay.

Jeromy Sonne: Unfortunately. That's in a wind down process right now. I'm no longer involved over there, but we achieved some things for sure. Seven figures in revenue and managed eight figures in ad spend over a couple of years. We shot up and grew, but ran into some scaling challenges. Couldn't get it across the line. And so, you know, from there, I was trying to figure out what was next.

Tim Rowe: What's your key takeaway when the dust settles on that? What's the thing you learned?

Jeromy Sonne: Ensure large contracts. That's all I'll say, I guess. Buy insurance.

Tim Rowe: Literally insure big contracts.

Jeromy Sonne: Yes, buy insurance on big contracts is the single biggest lesson. Without getting into the gory details, right? Basically, make sure your financial ship is in order and insurance is good if you're going to offer net terms, basically. But from there, I think that the biggest lesson now, how to best serve advertisers. There's just this fundamental question that has not been answered of how much do I spend on what platforms or channels? to get the best possible outcome. And that's such a simple question when you put it that way, but there's a lot behind it and it's extremely complicated. And I don't know if anybody can confidently answer that today, to be totally honest. I don't think that any single person can 100% there's people that are more right than others. So I became like obsessed with this, like idea of like. you know, holistic, probabilistic attribution and optimization. Ironic for a guy that failed business calc in college that I now spend most of my days like doing stats, basically running statistically significant relevant tests and designing them and executing them and figuring out the budget allocation between platforms and what correlates with success and everything. I guess I say correlate 500 times a day at this point, but yeah, that, that was really what became my obsession. So when I. Left that, I was like, I really want to chase that vision, but not just in the narrow sense of audio, but like overall, my goal was to be able to go to a brand and like confidently tell them like, Hey, these are the platforms you should use. And this is how much you should spend on each platform to get the best possible outcome, given what your goals are. Right. Trying to solve that simple question kind of basically became my, has become my whole career, uh, in one way or another. And you know, then it's just finding the right application. Uh, and so that's where day parts solutions has come from right off DSP. Um, we're going to help you, uh, dynamically allocate your budget across, uh, to do that. And then. We can all from there because one, the DSP model is extremely difficult. It is very, very hard.

Tim Rowe: What's difficult about it? What, what makes DSP businesses hard?

Jeromy Sonne: So Ari Paparo wrote a really good article. Gosh, I think it's seven or eight years ago or something like this, but he wrote a really good article about the challenges of the DSP model. And really it's because everything's built into markups percentage of spend and there's intense downward pressure. Um, on those markups, it's a high volume, low margin business, right? Psychologically, when you're working with a customer. Um, they are, they're not looking at your fee. They're looking at like the all in cost. Right. I mean, so it's very, very difficult to like have that conversation around we're netting out 5% or something. It's a constant conversation of why shouldn't I just go direct to these publishers? I can get a better rate. I can get better inventory, all of that. What value do you add? And our answer would be that we. dynamically allocate to give you the best possible spend, but then there's an education gap of trying to teach people why should you trust an MMM or a GeoLift study. And then when we have that conversation, then the next conversation is alignment issues. Like, Hey, if you're selling this inventory and you're getting variable rates on it, how are you not biased? And so. Kind of looked at that and said, that is an extremely difficult hurdle to get over. So we pivoted the model towards this AI media buyer that would sit on top of all of your channels and would directly integrate with various publishers through their self-serve platforms. Repositioned as we're agnostic, we're just helping you, we're advertiser side tech, where it's basically like an AI agent that will do your bidding for you to help you execute that.

Tim Rowe: Obviously the DSP model, a take on the, on the media spend tech fees, data, whatever. How are you making money on the AI media buying?

Jeromy Sonne: At the time was a percentage of media, but the difference was that it was a flat percentage across everything. So we weren't biased in one way or another. It wasn't dynamic, like, Oh, we get higher margins on display and lower margins on CTV or something like that.

Tim Rowe: Right.

Jeromy Sonne: Because it's flat, I don't care where you spend the money. I just care that you spend the money. And that went okay for a while. We still ran into that issue of the education gap where we spent a lot of time building out these MMM models and even lining up multiple MMMs that were open source against each other and aggregating the results and doing all sorts of things to try and build trust. And then at the end of the day, education gap still existed for the brands we were targeting, which were the performance brands. We could try and go upstream, but we've found that really around the 500 million a year in sales mark, we just didn't have the sophistication necessary yet to service enterprise customers. We were really tooled to do larger DTC brands and like regional CPGs. We added a lot of value there and I still think that, you know, we did and do. But, um. You know, kind of beyond that, right. It candidly, like it was the margin for error with those sorts of businesses is very, very low. You have to be extremely dialed in and there is just not a lot of tolerance. There's not a lot of slippage. Right. And that makes sense for them. Their margins are low, right? Like they're, they live and die by their better return on ad spend, their Google return on ad spend, that sort of thing. Right. And so we kind of like cornered ourselves basically into this like narrow sliver Of the market, it was kind of this like uphill battle all the time, right? It's not that we weren't, we were doing bad work or we weren't driving results. We were, it's just like, it just became like a real grind for everybody involved. What we did find though, is we had a couple of customers wander in that were not our ICP at the time that were more B2B SaaS companies. I remember looking back to some of the consulting I did and my marketing director job, I was like, you know, The return on ad spend is, is crazy high here for this, right? So the challenge for them, especially for B2B SaaS, cause there is very little marginal costs associated. It's over 90% gross margin. And then additionally, the challenges there were very different for them, right? But it was challenges that we were very. well set up to help with. So for them, basically they were only ever advertising on LinkedIn and Google search because that's about the only places they could confidently target the B2B company. And we're trying to reach. And so what we did was we said, look, between a combination of factors, whether it's a DMP approach, which we utilize and layering with probabilistic on top for the stuff where OOH and linear television, especially for like SMB focused, like a B2B companies. And we can actually add like quite a bit of value by unlocking these channels. Right? So if you're a DTC brand, right? Like the question is, Oh God, I'm going to say it. Curation.

Tim Rowe: He said it.

Jeromy Sonne: I said it, I hate myself for saying it. The point is for them, they have a sea of options and they're trying to narrow down to the efficient ones. It's the opposite conversation for B2B SaaS companies where they have almost no options. They desperately want more addressable media, but it just has to make sense. So between a hybrid approach of utilizing a DMP and then also layering on probabilistic methodologies to make some of the other stuff work. Now we're opening up entire new worlds for them. And because they're very laser focused on their numbers and we're typically working more with a VP of sales. Those folks are very dialed in on money in money out. And so when we can do things like run a holdout test inside of their pipeline and show the difference in conversion rate of running ABM advertising, um, to their leads, uh, it's very obvious, like win for them. And, um, you know, and because there's this like. higher margin for their customers, it's very easy to make a lot of these channels quite profitable for them. So it's less of a efficiency question and more of a scale question. And we're just better set up to handle that. And yeah, this ABM advertising works. You don't even have to ask me to go look it up. 70% of B2B marketers either want to or have to some degree implemented ABM ads because the conversion rates increases on the deals in their pipeline. It just goes way up. And that's all publicly available data and studies that people have done and stuff like that. It inherently works, but challenges. How do I activate it? Because most platforms don't have B2B targeting. How do I actually go and juggle all of these channels? B2B marketing teams are less well resourced than a lot of consumer because it's more of a sales focused organization. And so that's, that's a position, that's a place we're really able to help with. And the early, early results have been extremely promising. And we're actually in the middle of finally launching our self-serve solution here with Daypart this week. Yeah, we got our first couple of customers on the platform and raising capital. And we, I think we really found a great niche here that we can build off of and expand out to over time. So found that, that ICP customer that really needs what we're doing. And then as we come back to that system and get it more and more dialed in, then we can start to slowly grow back out. Because advertising isn't the sexiest industry in the world, despite what Mad Men would tell us. I do believe in it, though. It fuels capitalism. A advertising market that allows access to anybody and puts them on an even playing field with the ability to buy inventory at a fair rate and measure its results and optimize it is good for everybody. It's good for small businesses. It's good for consumers. It's good for publishers. I think that that's, that's really the goal here. And so B2B is the right place to start with that. Cause they've, they typically don't do a lot of advertising cause they haven't been able to make it make sense. And I think that we're, we're bridging that gap.

Tim Rowe: Listen, I love it when I could wind up a guest and let them rip because there's gold backed up in that. You just gave us an entire two startup journeys. You walk through product market fit AI. We, we, we explored a lot there. So I'd like to double click into a couple of those things. And one of them is it's in the name day part AI, but how do you think about AI as a part of the overall stack?

Jeromy Sonne: Yeah. So you have your AI maximalists that are like, AI is going to do everything. And we're all just going to basically be a bunch of feeder cows that wander around in the world while our AI mommies and daddies take care of us. Right. And then you have the other side, which is like AI is a total farce. It's basically crypto over again. No hate against crypto. I'm crypto. I know some great crypto people that are doing good work and I'm sure they're finding their way. I'm sure it will be part of the future of technology, but yeah, they kind of view it as like scam or some. a cheap crappy way we're going to have to deal with this technology that doesn't really work because corporate profit margins like it and it's not really that great. It's not revolutionary. I'm somewhere in the middle. I don't view it as it's going to eliminate tons of jobs. I view it more as it's going to free people up to do what people do best. Instead of us being all chained to our laptops, pushing buttons in various platforms, the AI will do that. And we can focus on the things where humans add value through like creativity and strategy and customer empathy and all of those things, especially in advertising. So where I view AI as being extremely useful is yeah, getting rid of like all the buttons clicking all of the rote monotonous tasks so people can do great work in advertising. And so I'm actually less bullish on like the, um, gen AI side of it. Right. I think that has a place for sure. I think that like, especially like, you know, taking a piece of ad creative and tweaking it 10% for like each person to like, you know, hyper personalize, like that's a great use case. Right. I think we're a long ways off of like AI coming up with like a full creative campaign, positioning, you know, all of the stuff that like, advertisers are great at. Um, you know, where I, where I do see it as like, it's going to be like a force multiplier for like media buyers and stuff. So you're basically going to have somebody that's more of a strategist. Somebody who's more of a media director, instead of having to manage 50 people or a hundred people, they're going to manage like two senior people that will then manage AI. And then those other jobs, I don't think they're going to go away. I think they're just going to shift more to the creative side of the house and find other places to To add value, I view it very much so as opening up human potential by getting rid of a lot of the monotonous stuff where we're not adding that much value. And so how we specifically use AI is number one, we're using AI to actually hook into your CRM and use that as a source of truth. So closed one deal equals purchase. If you want to think about how we've transitioned from working with a DTC brand. Lead equals website visit, that sort of thing. So one, we're using that to do a lot of data enrichment and, um, you know, understand where can we actually reach these leads? Um, you know, both in the deterministic sense of we were able to match it on this ad platform using one-to-one. Uh, and then also in the sense of I'm targeting small businesses, I want to use something like OOH, where there isn't inherently a one-to-one and say, probabilistically, all the OOH inventory that exists, what should I be buying? And then using AI to actually execute things like lift study test and run MMMs, be able to tell what is the impact of that OOH, which inventory is performing the best, what can I pay for that inventory, all of that optimization. The other place that we're using AI is to figure out what is that budget split between channels, ingesting all of the data, looking at probabilistic outcomes and then being able to go there. And then finally, I'm a big believer in one of the biggest value adds it can add is adding, adding value to like the user experience. Like if you've been in the tech industry and you've seen like, you know, some of the, how this stuff goes, you've worked with all sorts of folks in the organization. How many times have you gotten an email that's just like, Hey, can you pull a report for me with no additional context? And now you're scrambling around trying to figure out what they actually want. Having a chat interface for that, I think is huge. Right. And being able to say, okay, yeah, I'll jump into your data lake and I'll pull a report, answer specific questions about it, all of that. So I think it takes the burden off of the human to human communication in a way. Right. By 80, 20. Can 80% of that answer be done by AI? Yeah. And then like human just clarifies. So, uh, that's where I really think the user experience is big too. That and writing back to it, right? How do I, as a creative person, creative director or client, add context of the qualitative sites. Like we really want to avoid political content or something. And then the system being able to translate those insights into a quantitative outcome. actually deciding, okay, I understand that you are telling me that your best customers are X, even though that might not show up in the data. So I am going to make those adjustments now to target more X customers instead of just blindly chasing Y. Yeah. So I think that that's really one of the most interesting places that AI is going to help is translating those qualitative insights into quantitative execution. So. That's how we're using AI. It's not some unified system of AI at this point. Eventually one day we have very big ambitious plans to eventually build an inference model that can tell what people are going to buy before they buy it. Utilizing probabilistic signals and building some giant probabilistic predictive model using AI. Right now, it's in some discreet, interesting ways that just improve workflows and let people again, focus on, on what they're supposed to do. B2B markers specifically, you know. Click less buttons and less ad platforms and spend more time crafting content that is going to actually move the needle forward. Spend more time thinking about your brand and long-term strategy instead of just fiddling with LinkedIn ads or whatever today. So that, that's where. I think AI is going to shine in this industry. Love that.

Tim Rowe: I heard it showing up in strategy. I heard AI in planning, AI in buying, AI in measurement reporting, the ability to use natural language processing to communicate real time with the data. So it sounds like AI is, is definitely a big part of what you're working on. Jeromy, we're going to come back and do a little bit of news, but before we do that, give folks the Latin long, where do they learn more about Daypart? Where do they connect with you?

Jeromy Sonne: Yeah, go to daypart.ai and check us out. If it's something that you're interested in, would love to connect. You can always email me at js at daypart.ai or feel free to hit me up on Twitter at j-e-r-o-m-y-s-o-n-n-e. I'm pretty good on my answering DMS and always down to chat. Uh, don't be alarmed by my Twitter. I view it as my, uh, online water cooler. So I like to have a lot of fun and crack a lot of jokes and I don't, yeah, I don't take myself too seriously on there. Um, I think too many people take their public personas way too seriously. But for me, I just like to have a good time. And I think that resonates with a lot of folks that are kind of burnt out about yet another thought leader thing. Right? So that's, that's kind of how I've been. There is no strategy for what I do online, right? But I just like to have fun.

Tim Rowe: The strategy is no strategy.

Jeromy Sonne: I've seen that resonate. Yeah, the strategy is yes, absolutely.

Tim Rowe: Love it. We will make sure to link to all of that in the show notes and close by so you'll be able to get in touch with Jeromy and learn more. We'll be right back with a little bit of news. We've had a lot of interest in last week's conversation with Jonathan Moffie from Streamr AI and specifically how they're equipping DOOH publishers with generative AI tech that unlocks long tail demand from small and medium sized businesses, tier two and three agencies by making it easier than ever to generate TV quality spots with the power of AI. If you've ever lost a deal, because the brand or advertiser didn't have the right video creative, you know how much that sucks. So do not let it happen again. Visit sellmoredo.com. That's sellmoredoh.com to schedule time directly with me to discuss how Streamr AI can solve that problem for you once and for all. Add meaningful incremental revenue in 2025 by adding easy to use technology for your sales team, for your clients, to your already powerful network. Visit sellmoredoh.com today to get started. Now back to the show. Jeromy, for today's news, I pulled something special just for you. I pulled this recent SiriusXM report because I know audio is near and dear to your heart. It was a core part of the thesis at Decibel. I know it's a big part of the work that you've done. Pulled this report from SiriusXM talking about traditional linear TV being down 43% in the same period that audio consumption is up 183%. 75% of the US pop over the age of 12 regular, regular, 75% of the U S population over the age of 12 regular regular. Why is that so hard to say? 75% of the U S population regularly listens to digital audio podcast, streaming audio, all that 4.8 hours of consumption a day. Power users. It's intimate. It's direct to ears. Everyone loves it. Everyone knows it's high impact, but it's hard to connect the ecosystem. I've heard not having creative is a huge challenge to not running audio. Curious your thoughts overall. Does audio deserve more love? It feels like audio deserves more love. This is coming from the OOH guy who feels like the perpetual underdog.

Jeromy Sonne: I do love audio still. It has a near dear place to my heart. I am very bullish on audio. You know, Decibels Adventure wasn't the right way, I think, to highlight the value that audio has. But Jeff Green from Trade Dust said basically it's one of the biggest mismatched areas of spend versus user attention. And I think a big part of that, yes, creative folks learning how to do good creative for audio is definitely. a skill that needs to be leveled up for a lot of creative folks, but they'll get there for sure. Radio is huge and still actually has a lot of value for a long time, and there's a lot of lessons learned there, but I think people are addicted to the quips, and that said, things are starting to change. I'm bullish long-term on audio. It's just going to take some tooling, take some understanding, and then we will eventually So I'm, I'm long-term bullish on audio. Short-term, I think we still got some work to do, but it will get there. And I'm, I'm with SiriusXM, it being an extremely valuable platform. And so I hope folks recognize that more and more. And the earlier you do recognize it, the better value you're going to be able to extract from it as an advertiser on the buy side.

Tim Rowe: As a DOOH guy, I think that there's a great story to be told of DOOH, audio on the go moments. It's an intimate connection there. Jeromy, I can't thank you enough for being here. I appreciate the conversation. I appreciate the transparency.

Jeromy Sonne: Appreciate you letting me ramble. I'm always down to talk my head off and say a bunch of dumb things. So hopefully folks got value out of this and yeah, always nice to chat. Always nice to chat.

Tim Rowe: Love it. We will make sure to link out to all the places to connect, learn more, follow you on Twitter until next time.

Jeromy Sonne: See ya. Excellent. Thank you. Appreciate you having me again.

Tim Rowe: Thanks Jeromy.