Tim Rowe, Host:
Welcome to Radically Candid, a Cognition podcast that takes you behind the scenes with the people, personalities, and perspectives shaping how we think about streaming TV and how we approach solving challenges for agencies who are trying to build an owned and operated ad tech stack. My name is Tim Rowe and today's conversation is with Colin Davis, CTO here at Cognition. And we talk all about all things dynamic, creative, what it takes to ingest big inventory feeds, lots of data, how to create personalized data-driven creative at scale, how Colin thinks about building a team, not only a team of devs and engineers, but also how that team scales and what that impact looks like as it grows. So sit back, enjoy this conversation with Colin Davis, Chief Technology Officer here at Cognition. Colin, thank you for taking some time out of your very busy schedule. Looking forward to today's conversation to talk all things technology and technical. You're our CTO. What does that mean? What does a CTO do at an ad tech company?

Collin Davis, CTO: Sure. I mean, CTO, for those who don't know, stands for Chief Technology Officer. Specifically, my role is surrounding the orchestration of development resources with product teams in order to make sure we're executing on building features that our customers want. I see my role more so as actually less technical, not like hands-on code, but more the kind of orchestrator or the conductor of what are the business requirements, how do we prioritize those on our development roadmap, and how does that translate into certain outcomes. So my day-to-day is really kind of meeting with the team every morning, meeting with the product team, understanding their wants and needs, what clients are saying, and then ensuring we're kind of taking the best approach to implement any of these new features, or architecting or designing a system in a way that is scalable and efficient. And or maybe some skunks works projects where we're actually integrating or testing with a i and other new features that can help make our platform better.

Tim Rowe, Host: You said the magic word you said a i so we'll come back to i know there's already a i in the platform that doesn't get talked about enough so we're gonna talk about it together and ultimately. Today's conversation will lead into what you see driving the direction for ad tech, especially as it relates to streaming TV. But you said something important there, and I'd like to double click into it. You talked about building features our customers want. Seems like too often it can become easy to build the feature that we think is fun or is kind of flashy, shiny. How do you get to like the core of what problems our customers have and the features that need to be built?

Collin Davis, CTO: Well, a lot of it comes from the product team and the account managers. So let's start with the account managers. They're hearing from clients on a day-to-day or regular basis. They may stumble into something going over a demo with them or clients just say, hey, be really cool if I could do this. So then it funnels to the product team who kind of specs it out and says to my team, hey, here's what we want to do. Here's the why. Now tell me how or figure out how to do it. I always like to kind of describe it that way. Development is really the how of something. The product team is the why when they talk to these users or these account managers. So We have a very rapid development process. We do really weekly sprints. And essentially, these one-week sprints are kind of our mini marathons to run through a bunch of features, get them out, and get them into the platform as quick as possible. I'm a huge believer in speed. I respond quickly to emails, texts, Slack messages. I provide feedback to developers as soon as possible. Like, speed is everything. So we're a very fast development team. We put a lot of emphasis on that while not sacrificing quality or long-term decision-making. So while we may spend a lot of time thinking about a problem and how to solve it, when we execute, we do it very quickly. So the question about, you know, how do we field in those customer requirements? It's really a very fast, regular feedback loop between account managers, product managers, and then the development team.

Tim Rowe, Host: And when this was going to be part of our conversation that I had originally planned for the end, but since we're already here, let's tie it in, then we'll weave this all full circle. How do you think about that then? How do you think about structuring your team with the problems that you solve and shipping product? How do you think about team structure and actually triaging these to work as quickly as you do to ship, to iterate, optimize? How do you think about building the team?

Collin Davis, CTO: So from a talent and a human resource standpoint, I've been hiring folks both remotely and in person for over 20 years now. Wow. So my experience in doing that helps me find the right people that are not only a culture fit, but also and when I say culture fit for me, like attitude is everything. OK, so I really like problem solvers. I don't like folks who say no to something. Um, and kind of shoot it down because it's not the way we do it. I'm a big fan of change disruption and change is how you progress. That's how you advance and speed really plays into that. We have to be willing to take chances. 1 of the core missions of cognition is to dare to fail and to try new things. So. I find people who are always like really love or are passionate about problem solving. And then when you get down to their skill sets, I mean, you know what kind of languages you're building in. So that's kind of the second piece of it. Are they the right culture fit? Do they have the right attitude? Are they true problem solvers? And then what are the languages that they know? Make sure that the languages are also applicable to the things that we're building or using. And we use a number of different languages across our platform and our different products. So having someone who's what we call a full stack developer to me is the most valuable thing versus someone who's just specialized in this one other thing. If we have a need for a specialized person, we can contract with them. So we can bring folks in to help with certain problems that they've solved before and we can spend a couple of weeks doing that and then hand off the learnings from that to one of our full time kind of core developers as well. But again, it's really about that attitude and then it's about those. those skills in terms of languages. And then from a process standpoint, there is a collaboration loop within our team. One of our software development managers, Max Kramer, who joined us a few months ago, has done a great job of implementing that. So he's been both like a code mentor, and really helping the team review and stick to best practices so that everyone is consistent and continually on the same page and kind of looking at or reviewing or working on different pieces of code that they may not see if they were just kind of in a silo working by themselves.

Tim Rowe, Host: That's huge. And Max, coming by way of one of the acquisitions that folks have probably heard us talk about on a previous episode like this, That makes a ton of sense. And thank you for walking us through it. I think it actually lays the groundwork even better for the conversation. The North Star I'd like to follow maybe for the duration of this is what do you see driving the direction of streaming TV ad tech?

Collin Davis, CTO: Well, AI is obviously a buzzword we all know very well in every industry.

Tim Rowe, Host: Sure.

Collin Davis, CTO: You also have RPA, which is Robotic Process Automation, which to me could be another spinoff or version of, quote, AI. At the end of the day, we're talking about automations to create efficiencies. So at Cognition, we're really focused on our automations that are pushing and pulling data. This can be first-party data, it could be vehicle data, it could be reporting data. We spend a lot of time ingesting a lot of data and then reporting back on it and helping our customers make better decisions so you know let's just say five years ago you were focused more on like analysts humans going through this data and kind of coming up with some. Or asking the right questions and getting certain answers and then reporting back to the clients will now a I can do that you can train it to look for anomalies or to help with. Making suggestions or recommendations and then part 2 of that is you can then ask an AI or another system to act on those recommendations. So kind of like the point would be, hey, I'm running a campaign. This audience isn't doing that well. My click rates are low. What can I do to change it? And AI can find those anomalies and those issues, and then it can act on it and say, oh, well, let's try this other audience or let's change up our targeting to get better results and then monitor it and iterate over and over. So. The future for us with streaming TV is being everywhere. It is measuring our attribution. So like streaming TV has been tough to measure. I know Carson and Michael have probably spoken to this already, but there are now ways that we can literally paint the full picture of the life cycle of the customer journey. Streaming TV used to rely upon QR codes for you to scan in order to attribute, you know, knowing if someone really watched that ad. We have other technologies in place now, including our hat or headless analytics tag, which allows us to match our impressions of these ads to folks who also visited the website and kind of close the loop or close the gap on that person's visit and what they're interacting with, where they're coming from, so on and so forth.

Tim Rowe, Host: I always felt like QR was, it was almost a misrepresentation of what was going on, because it was always going to be a subset of a subset of a subset of the total audience that was exposed to it, saw it, decided that they were intrigued enough to pull out their phone, their phone could support it, they had enough connectivity, and then they actually went and did the thing. We ran into similar challenges, obviously, with the out-of-home and billboards, people wanting to measure it. but now you've seen QR kind of make a resurgence in streaming TV, but more intentionally as like a sidebar call to action or on some of the pause ads and different ad formats that are coming out where it makes more sense. So it's a great mechanism if we understand kind of who it's designed for and what it's designed to do versus as a way to measure the entire channel's performance. You said there's better ways to do that. How do you think are better ways today that that exists that maybe didn't during the earlier days of streaming TV, though, it seems we're still in the earliest of days.

Collin Davis, CTO: Yeah, I mean, there's always new ways to do it. It could depend on the platform you're advertising on. So for example, with Amazon, they offer some ad experiences that are native to their platform, like Prime Video, where you don't have to scan a QR code. You can actually just click a button on your remote to say, I'm interested, and it'll send a lead essentially to whatever your Amazon account email is. So different platforms offer you different capabilities to improve the experience. But nothing is really uniform. The QR code is probably one of the only uniform ways to do it. Like I can put a QR code on top of my ad and hope that you scan it. But I think to your point, people aren't necessarily going to scan it for whatever reason. Scan rates are still pretty low. So what's the other way to measure that someone saw that ad and then maybe they went on Google and did something to look you up, right? they found you just by a different channel. That's where our impression tracking really comes into play. So we talk about HAT, which is Headless Analytics. We have some custom trackers that we built. And essentially, we have been whitelisted and approved on all these different platforms. So we can close that loop. We can see when the ad was seen. And then we can basically connect that view or that impression to the person that also happened to visit the dealership or the website of the client that we're serving. And that's how we basically close that loop. Being an ad server is a huge part of that. So Cognition is an approved ad server on Amazon. Our ad serving capabilities also work on many other DSPs like the trade desk and basis and so on. So again, we've got all these pieces of technology that are very useful to help you measure the effectiveness of your ads that were definitely not available before. So it's a combination of technologies. How it all works is pretty fascinating. And one of the more exciting parts about the job is working on and building those capabilities. Which parts are most exciting? Well, just the fact that we can see what you were doing before you landed and then after. So if we've been tracking a website for 30 days, and then we start a campaign on day 31, if you see that ad, we can say, oh, 14 days ago, before we ran this campaign, Tim had been on the website and looked at Ram trucks. Right. But it looks like he actually saw the ad, went to Google and searched, then landed on the website. So this idea of like last click or, you know, giving one person the credits, it's just not, it's not a real expectation here. The expectation or the, the, the right way to state it is that it's an assist. So we can show how we assist or that we influence that process and attribution to us is kind of many of those touch points and being able to measure them and report back to you that they happened.

Tim Rowe, Host: So if my understanding is correct, it's kind of those two pieces working together. It's the ability to serve the ad and have the pixel on the website to observe what happens kind of down funnel. Is that correct?

Collin Davis, CTO: Exactly. So when we're setting up our ad tags within our ad server, There's a lot of things that are working in concert together from like the creative to the targeting criteria. But for the tracking mechanism, we are using the same database where we're picking up all this tracking information from the website, and we are matching it to any tracking information we also collect inside of the ad tag. Right. We kind of have both tracking scripts in two places. And one is collecting the data on the ad serving side. The other one's collecting data on the website side. So that is how we kind of bridge that gap and we tie your visits or your interactions together.

Tim Rowe, Host: Wow. Okay, cool. Thank you again for the refresher and just added content for the marketing engine, as it were, about how that actually comes together. I want to come back because I feel like especially our operations folks, any technology leaders that might be listening to this, thinking about how do we integrate, I want to come back to the automation piece and what you're thinking about from an automation standpoint, because It is a hugely critical element to making all of this work, which is automating things that would otherwise be done in a pretty manually intensive way. Can you maybe give us some examples of throwaway things that people do every day that are pain in the butt that we're automating and kind of solving for? I think that there's a lot of interesting use cases people would probably be intrigued to hear.

Collin Davis, CTO: Yeah, so a quick kind of backstory is when I founded Cartender, which Cognition acquired in January of 2024, my whole thesis for Cartender was let's remove the monotony from repetitive tasks in the video editing world.

Tim Rowe, Host: Now Creative Studio.

Collin Davis, CTO: Yes, now known as creative studio by cognition and so creative studios whole mission was to basically take that repetitive task like updating your offers on vehicle leases or finance offers or whatever every month. and replacing that with automation or technology that could do the heavy lifting for you. So Creative Studio is one of those automations that we have that's really, really cool because it takes hours or even days of work and turns it into minutes. It's a highly scalable system. It can take that data from any source and it can automatically update images, videos, different sizes, you name it. So it's a really, really great tool for our clients who are trying to automate their creative process. Then we go a step further. We implement our ad serving layer on top of it. So not only can you auto-generate your creative, you can also automatically update your ad tags with that creative, which means you never have to go into the DSP to modify your campaign to change your creative. It's all automatically done for you. And then we have another layer to it. And the other layer is how we correlate or we really marry the idea of targeting strategy to creative strategy. So if you're targeting people who are in market for SUVs, it's highly likely that your creative strategy is to show them SUVs. So we have also built a lot of fantastic automation, and it's one of our unique kind of advantages to the Cognition platform is that we can take your targeting strategy and have it inform the creative strategy, which automatically turns into creative that's generated, which automatically updates your ad tag. So another loop that we kind of close the gap on when we talk about How are we targeting or positioning our campaigns? How are we spending our money? The creative piece almost becomes an afterthought. We just automate the entire creative process, which is also the most difficult process for a lot of our clients. It takes them a long time to generate creative. We have a full kind of end-to-end solution to give them everything they need in one platform. The other thing to think about, Tim, is that a lot of DSPs, they don't have creative capabilities built in, except for, hey, upload this MP4 file or upload this image. So when you really layer in our idea of creative strategies correlating with targeting strategies, and you add our ad serving element to it, It is a full complement of features that makes everyone's life really easy and really efficient. We've literally removed a lot of those kind of roadblocks or those time delays that often occur within this process.

Tim Rowe, Host: And the piece that stands out to me is that creative is oftentimes the most deliberated piece of the process, too. It's subjective. Everyone has an opinion and by implementing a scaled automated way where you have control, you have control over the inputs, you're connecting to a dynamic, always changing inventory feed because creative is oftentimes so deliberated by having those mechanisms in place. It actually allows you to have higher level conversations about the business. We're not wasting as much time in this creative process. Now we're talking about outcomes and how do we create different campaigns designed for different business objectives? And that's a totally different relationship than, hey, can you get me an approval on your creative so we can get it live?

Collin Davis, CTO: Yeah, I mean, when you talk about the business objectives, if you're talking about auto, really, or any industry, I mean, we know it very well. So right out of the gate, you're talking to people that know exactly what you're going to be doing because we've done it before. So we know the creative strategies that exist. We have all those playbooks. The same is true for any of our media strategies and campaigns that we deploy. The other really important thing is time is money. If I've got to wait two weeks for creative to launch my campaign, I just missed two weeks of getting in front of my customers, right? When those offers come out every month, you better have my video updated and ready to go and in front of people very quickly. So our automation solve for that pain point, which is you're not waiting for creative. We have it. It's available within minutes. It's updated automatically in your campaign. You literally don't miss a beat. I mean, you know, within 12 hours, I would say, or less, you're going to have that updated creative. This includes mid-month creative changes. If the car sells or if you need to change a photo or change your offer, whatever the reason, those changes occur near real time. It's a very fast process on our side. Another thing to consider that really comes to mind with our automations is when you're creating a piece of creative for your ad, whether it's a display ad or even a video ad, you often wonder like, OK, well, like, what can I change about it? How's it going to look? At the end of the day, or how we begin the day, is creative is flexible. You can do whatever you want. We'll figure out the data piece. And what's really cool about the Cognition Creative Studio and all of our platforms is we can take in any type of data source. It could be inventory data. It could be RO service data. It could be sales data. We can use that to create content. That really speaks volumes to not just advertising content, but really any kind of marketing content you could think of. So for example, you know, in your service data, in your RO data, someone got a brake repair done, but they refused another piece of service. Well, if that's noted, if that's in the data that you're sending us, we can actually generate content around that to try and get that person back in. That could be an image, that could be in video. I mean, there's so many different ways that we can automate with our tools to help you create dynamic messaging that you can use in a number of different ways, not just advertising.

Tim Rowe, Host: See, I'm learning things. And in fact, I'm playing with Creative Studio for Cognition Marketing because there's also AI features in there to create voiceovers from text-based scripts, plug-and-play music beds. There's a lot of utility even just for my uses. It seems like everything, if there's a convergence of the conversations, leading up to this, Colin, it's that there's a funneling effect towards the activation analytics feedback loop and being able to propel that as quickly and as viciously as possible with all of this stuff. And the creative piece being kind of essential to that. I wanna talk maybe about the decision-making that ad serving enables for DSP partners, knowing that that's something that we're focusing on later this year and starting to use some of the really, I think, sophisticated decision-making that goes into automotive advertising, display and video, the decisioning that goes into those things and how those tools can be used by maybe traditional ad tech partners. Is that something that you'd be comfortable exploring?

Collin Davis, CTO: Yeah, so I think of it this way. A lot of the decisions we make today are premeditated.

Tim Rowe, Host: Okay.

Collin Davis, CTO: We know that we want to target certain people. We know that that correlates to a creative strategy based on what we want to show them. So we know who, we know what. But then when you think about ad surveying and kind of more of the dynamic pieces that you can take advantage of, what if your assumptions were wrong? What if our system learns that a certain piece of creative isn't doing that well and it should try something different? we're working towards those models that can make those decisions. So you don't have to, you actually don't have to wait a month or a week to measure your results and say, ah, it didn't do that well. Let's try something different. We can act upon this data. We can model it and we can iterate in near real time in order to constantly make those improvements so that by the end of the month, you can see that we made those changes, right? Assuming they're within your policy that you allow and give you better results as a result of those changes. So those are the types of things that we're thinking of long term. Being an ad server gives us access to data that we normally would not have in a premeditated kind of assumption thought process. We are retrieving this data and we're able to make decisions about the creative and even about the campaign that wouldn't otherwise be possible.

Tim Rowe, Host: And how does that benefit maybe a DSP partner who would be considering partnering with Cognition?

Collin Davis, CTO: Well, because we are the missing piece of the creative in this example, right, we're not just ad serving like a typical static image, we're ad serving a piece of creative that's changing. That means campaigns perform better. That means DSP platforms get more dollars in order to kind of be spent towards those things that are working better. It also allows us to use their platforms in ways that maybe they haven't seen where we're dynamically changing audiences or bids or what have you in order to optimize the best use of that platform. And again, from an ad serving perspective, we're opening doors to industries like automotive, where there is a special requirement and how you do things, especially in creative, where if a DSP wants to be known as someone who supports an automotive-centric company like Cognition and their ad server, it's going to bring more dollars through the door.

Tim Rowe, Host: Cool, I'm sold, but it can't all be sunshine and rainbows. Where are you blocked? What challenges exist for you from a hiring standpoint, a development standpoint? What are the biggest challenges and blockers that you face in developing all of this great tech?

Collin Davis, CTO: Building a team that embraces collaboration the way that we do is a big one. Finding more of those culture fits is really challenging. It takes a long time. And sometimes you gotta test the water before you really know. The technology that we've built, I mean, there's a learning curve. So when you're hiring people, it's a ramp up period that takes a little bit of time, but over time, if all the pieces fit, then things run smoothly. As I look at the broader landscape of technology in ad tech, integrating with many different platforms that have many different requirements is the biggest challenge, not just from campaign creation, but from reporting, from pushing data in to create audiences from first party data. looking at our own data warehouse and how we make that as flexible as possible to connect to all these different platforms. Because more are popping up and there's more work to be done to basically support them so that you can use your data in any of these platforms that we work with, and our ad server can also talk to or work with those platforms that makes it as compatible as possible. The third thing, AI changes so fast that you could land on a piece of AI today and then next week there's something better. So really staying on top of that is the biggest challenge that I've ever seen in my career. I'm used to speed, but this is a different type of speed.

Tim Rowe, Host: Yeah.

Collin Davis, CTO: It makes it almost impossible. And I would say you get kind of AI fatigue because you're looking at so much of this, you're following it. You're trying to put a stake in the ground to say, OK, we're going to commit to this and then something else better comes. So finding a way to work with different AI technologies and being very flexible in your experimentation is a really important part of our process as well. So we spend a lot of time looking at this stuff. and trying different things. We're also a part of a lot of alpha or beta programs with our partners, which is a really cool way to innovate and do different things. And we're really at the forefront of that. Amazon announced at their latest unboxed event something called the Amazon Data Manager, or ADM. We were one of the first to test that and essentially send it all of our data, right, get it loaded in, making sure it was syndicated out to the different systems that use it. Partners like working with us because of things like that. We help them make their products better. We're really collaborative. And I think those kinds of partnerships really benefit everyone, right? A rising tide lifts all boats in that example.

Tim Rowe, Host: Let me ask for a hot take. You mentioned first party data in there, clean rooms. Is it all worth the hype?

Collin Davis, CTO: Yeah, I mean, especially if you're trying to do some retargeting that a website pixel alone can't capture. So think about it this way, a website pixel will capture someone who came to your website, but you don't know a lot about them. If you have the first party data, you paint a much clearer picture and that informs you on your targeting, your creative strategy. It allows you to make more accurate decisions about how you want to reach that person. What are you most excited about? What are you looking forward to? I'm most excited about our ad server tech. Again, it's close to my heart in terms of my whole career in automating creative. I did a lot of it with video. We've done a lot of it with images. So Creative Studio at Cognition is at the forefront of that. I'm really excited about the ad serving aspect of it because it's not just about creating something that someone sees it's about receiving information about that person and also making decisions on that creative based on that data that you're collecting if you're running an mp4 file which is a video file or a jpeg image It's a static piece of creative. It doesn't give you back anything, per se. With our ad-serving technology, we're serving images and videos and, like, really HTML-based ads. There's a lot more you can do from a collecting standpoint, and I'm really excited about what we can collect and what we can do with that information to improve the ad experience. The other thing that I'm always excited about, but also kind of getting fatigued on is the AI. So there's just so much you can do in that world. There's a lot of innovation happening there. How we use that in our analysis and our campaign setup, all of that ties back to all that automation work. We're constantly kind of improving and iterating and making our processes more efficient. So our time to market is just getting shorter and shorter with our campaigns and just everything that we do.

Tim Rowe, Host: Awesome. Colin, we'll cut it there, but there's obviously a lot more we can talk about, and I look forward to doing more of these with you throughout the year. We'll get more of the team involved. I would love to hear from maybe more of your team as well in a future installment. Can't thank you enough for sharing everything that you have today. All right. Thanks, Tim.