S2
Episode 9
Driving Performance With Contextual Advertising
About This Episode
We discuss current contextual advertising technology and the bright future of contextual advertising in the programmatic landscape.
Yang Han | CTO, StackAdapt
Vitaly Pecherskiy | CEO, StackAdapt
Ned Dimitrov | VP of Data Science, StackAdapt
Transcript
Episode Introduction (00:00:00)
Contextual advertising really gives marketers more of a peace of mind about what data they use when they do targeting and topic of contextual advertising is especially relevant now because it’s gotten to a point where it can drive on par or better results in audience targeting. So naturally, it’s in the spotlight, where it can deliver results and also address the topic of privacy.
How Agencies Thrive Introduction (00:00:27)
Curious to know what industry-leading marketers are looking to achieve and the ever-evolving digital landscape. The How Agencies Thrive podcast by StackAdapt is dedicated to helping the new breed of forward-thinking savvy, lean and mean marketers win in the rapidly evolving digital landscape. Time to thrive.
Matt (00:00:55)
Hey, everyone, welcome to our first episode of 2022. My name is Matt Evered. I’m the host of the How Agencies Thrive podcast. And I’m also the Education and Development Manager at StackAdapt. To kick off the new year, I’m joined by three guests from within the StackAdapt organization. And let me tell you, I have definitely learned a lot from them over the years. On Deck. We’ve got Ned Dimitrov, the VP of data science, Yang Han, our CTO. And, of course, Vitaly Pecherskiy, StackAdapt coo. Now, for anyone who’s been following our company for any amount of time, I’m sure you’ve heard from at least one of our guests in the form of a blog post a webinar or any other thought leadership content that StackAdapt has put out. But for those who haven’t, I think the best way to kickstart the episode would be to go around the table here and get each of our guests to briefly introduce themselves. Give us a look at their experience and how long they’ve been in the programmatic industry. So starting with you, Ned, if you could, tell us a little bit about yourself. We’ll get started from there.
Ned (00:01:55)
Thanks, Matt. Hi, everybody. My name is Ned. I’m VP of data science at StackAdapt. I joined StackAdapt in 2017. And I’ve been heading the AI and data science effort since then. So what that means is basically all the decisions that are involved in creating performant advertising campaigns. And there’s many decisions that have to be automated in programmatic advertising, because there’s just too much data for an individual to sift through it and make the decision. So that’s why we need to automate computers to do these things. And so those are the things that I concentrate on here at StackAdapt. Prior to joining StackAdapt, I was a professor for about a decade. I worked in two universities. One was the Naval Postgraduate School in Monterey, California. So University inside of Department of Defense in the US, where many of the students are mid-career military officers who later go on to be analyst for the Pentagon. And, after that, I worked at the University of Texas at Austin, which is one of the top engineering schools in the union in the United States, before joining StackAdapt.
Matt (00:02:53)
Fantastic. Thanks so much, Ned. Yang, over to you to tell us a little bit about yourself.
Yang (00:02:58)
Yeah, I helped co-found StackAdapt. Initially, I built the initial platform. And now, obviously, it has grown to encompass a lot of different teams and departments, including platform engineering, backhand, and infrastructure engineering, data team undernet, as well as security support QA, so a lot of different functions to help the team innovate and function in terms of, you know, our, our product and the day to day processes. Yeah, I’d say my expertise lies more on the backend side initially, you know, I used to build financial software as well as mobile app software. But yeah, now day to day, you know, we have to cover all the different areas to make sure that you know, we continuously innovate as a platform.
Matt (00:03:47)
Thanks so much. Now, Vitaly, you were actually on the first episode of the season. So I think we could probably do a quick refresh for our guests who maybe didn’t tune in at that point. But just tell us a little bit about yourself, a little bit about your history in programmatic.
Vitaly (00:04:02)
Thanks, Matt. So unlike Ned and Yang, I don’t actually come from an engineering background. So I studied finance at the university. And I ended up in tech pretty pretty much by accident. So in my first job, I was an account manager at an ad tech company. So there I was doing a lot of performance based advertising, primarily for game developers, actually, but that was even before mobile gaming, so even that was about 11 years ago. So since then, obviously, the space has evolved so much. And I had the privilege to see it grow from very rudimentary to where it is now. So prior to launching StackAdapt, I had extensive experience working with marketers and really understanding their challenges. So I really think of myself now as a person who connects the dots across the company to find new ways we can deliver value to our clients, and just new ways to grow.
Matt (00:04:59)
Thanks so much, Vitaly. Now today’s episode is definitely a very exciting one. Because back in 2021, when we were at the drawing board, you know, looking for topics to kickstart the New Year with, it was pretty obvious that there was one that we should be focusing on in the new year, and that was contextual advertising. So, without further ado, in this episode, we’re going to do a deep dive into the future of contextual advertising and uncover how it’s paving the way for a bright future in the programmatic landscape. You know, we’ll look at available technology, industry trends and other things to look out for in the new year. As a brief housekeeping note, you know, while I will be here facilitating this fireside chat, Vitaly is also going to be leading the charge. So as always, Ned, Yang, Vitaly, we’re very excited to have you on this episode. And let’s get started. Starting out, I’d love to hear from everyone here and get some initial thoughts on contextual advertising and really ask the question of, from the perspective of privacy. Why is contextual advertising more important now than ever?
Yang (00:06:06)
So basically, there’s been a number of changes in the industry in recent years. So this includes both government legislation changes as well as technology changes. So when it comes to government regulation, there has been introductions of new laws such as GDPR, in Europe, in which you must get client opt-in, in order to store as well as access user data. And this includes, you know, all the historical user data that ad tech is used to using in order to perform their accurate behavioural targeting, as well as for different platforms to share data amongst themselves. Another major change recently is with technology changes, and these includes the deprecation of cookies, starting with both so far is browser, and Mozilla’s Firefox. So historically, 3rd-party cookies allowed platforms to easily share browser-related data across the board and synchronize them. But with the deprecation of this, these companies have had to find out tentative solutions, such as universal identifiers. Chrome has also painted that great third party cookies in the future. So with these replacements, unfortunately, typically, they will not scale as well as the historical 3rd-party cookies. So there will be access to less data overall in that is that you can use for user based targeting. In addition, this data may also get less and less accurate, as more countries will attempt to also model data in order to supplement the loss of the scale. However, there’s also there’s also a push in the industry to get more accurate data through by synchronizing a CRM-related data when it comes to universal identifiers. So overall, in general, why is contextual more important? Well, because now you have two worlds when it comes to advertising one world where you have access to historical user data, and you can perform user-based targeting. In another world where you do not have additional information on the user, all you know is, you know what page and device the user is currently on. But no data prior to that, and no data from other platforms. So with this, you’re limited to the information on what is currently on the page. However, the page typically also contains a lot of rich information contextually, if you are able to understand exactly what the user is reading about exactly what their state of mind is in. So if you could leverage this as much as possible, and also take into consideration all the different pages a campaign is able to look at an overall performance of the campaign. You can devise very strategic optimizations and targeting solutions in which you can perform very well and reach your campaign objectives.
Vitaly (00:09:11)
To add to Yang’s point, aside from these operational challenges, there are obvious challenges related to risk. So I was just reading a new study published on Digiday. That talks about the recent study that came out where a company evaluated how many of top publishers actually drop cookie before consent, and they concluded that about 92% of publishers drop at least one cookie before consent. That’s obviously quite problematic. The study doesn’t talk about the brand side so it’s unclear how many brands or advertisers do something like that, but I suspect quite a lot. And I also suspect that it’s not always done intentionally as well, obviously, with the study, we have to take it with a grain of salt. But nonetheless, it’s pretty apparent that right now, the industry doesn’t have a good grasp on on how to actually implement legislation using technology. There’s a lot of challenges related to its implementation, because ultimately it it’s up to people that you have on your team to handle this implementation. And very few people know how to do it right, then there’s a question of me maintaining it all around, it’s difficult. And it’s, it’s becoming apparent that if it was so easy to implement, it would have been done already. So contextual advertising, it really gives marketers more of a peace of mind about what data they use when they do targeting. And and I think the topic of contextual advertising is, is especially relevant now. Because it’s, it’s gotten to a point where it can drive on par or better results in audience targeting. So naturally, it’s in the spotlight, where it can deliver results and also address the topic of privacy.
Matt (00:11:05)
So the second question that I have for everyone here is with contextual advertising, I think the general sentiment is that it’s incredibly promising. It’s a great solution for the future of programmatic. So with that in mind, why do you all think that this method hasn’t been used in the past? And why has it had sort of this rapid turnaround of being an emerging technology, now that we’re in 2022?
Ned (00:11:28)
I think contextual has been a classic type of advertising since basically the beginning of advertising. And the reason for that is that a natural way to place your ad is around relevant content to that particular message that you’re trying to deliver to the user. But the way that contextual has been done in programmatic historically has been a little bit programmatic. So let’s take a look at the two classic ways of doing this. So one way is through categories of contexts. So Bing has an entire category tree of contexts, that publisher could say this article is in this category, this article is in this category. And then the advertiser would sort of place their ad in all articles that are labeled as being in part of the particular ad category that they selected. So this is programmatic for a number of reasons. So first, there is an incentive from the publisher and to place the article in as many categories as possible. And this is because they want to get as many advertisers bidding on the article as possible. And so that’s one problem with it. Another problem is that the categories are just fundamentally not very specific. I mean, think about even the biggest tree that you can imagine the biggest category tree, what would it have, like 1000, nodes, like 2000 nodes. And if you think about the large variety of advertisers that are out there, there’s just way more advertisers and more products and more contexts that can be defined, then can be fit into a single category in being exhaustive is just a losing game. The second major way that contextual advertising has been done in the past is through key phrase matching. So in key phrase matching, the advertiser specifies a set of phrases, maybe 100, maybe 400. And the machine basically scans through articles that contain those phrases. And if the article contains that phrase, then the advertisement might be placed on that article. So this is probably problematic, because defining the context is very difficult. It’s very, very difficult for an advertiser to collect these 400 potential phrases that they want to use to identify the articles that they want to their ad to appear in, is very time-consuming, it’s very error-prone. And ultimately, you might include a word like Apple, let’s say, in your list of phrases, but that’ll put your advertisement both in context with Apple, the company and digital contexts and apples that people eat. And so the key phrases really don’t capture the meaning the context that the advertiser wants their article to appear in. So this is one of the major reasons why contextual in the recent past, hasn’t been seeing as much play as behavioural advertising.
Yang (00:14:21)
Yeah, there’s another problem with the static lists method of targeting where, you know, the context of a certain topic may change over time. So for example, if I want to target electric cars, well, new models of electric cars are being released pretty much every quarter now. So traditionally, client would have to manually modify their targeting list to add new models of electric cars constantly. But if the system were to know that you’re trying to target electric cars, then Ideally, you would want a system where it would automatically know what models of electric cars are out there, even when new ones are introduced. So you’ll automatically be able to be, you know, timeless, essentially, as campaign runs over, you know, a duration of time, especially if the campaign is a very long campaign, or if it’s a campaign that you want to run continuously. It requires a lot of manual effort to keep these up to date, especially when things change.
Matt (00:15:30)
Thanks so much, Yang. Thanks so much, Ned. There was something you actually touched upon, you talked about kind of these classic methods. And there was another question that I had here, which talked about these classic methods and sort of from then to now, what’s sort of new and emerging with this technology, and what have been the major changes that have happened to make contextual advertising a lot more viable today for programmatic campaigns.
Ned (00:15:58)
So what makes contextual advertising different today than it was a few years ago is basically the application of many new AI technologies. So computational power has gotten a lot larger. And a lot more tools have been developed to do natural language processing, to basically capture the meaning of things as opposed to their raw string text. And that’s what makes contextual advertising very exciting today, because now it’s possible with just a few phrases to capture the meaning that the advertiser has in mind for their content. So just going off of the example that Yang had in terms of electric cars, maybe the person puts in things like electric car charger or electric car, and then the system automatically identifies things like Tesla, Audi EQC, or Mercedes EQC, the Audi e-tron, the Porsche Tyco, so on and so forth, and then places your ad, in all of these additional pages, that you didn’t have to go and research yourself, it just the AI automatically identified that this is what you meant, and filled in all the details for you. And thus, was able to find you fantastic contacts, without you having to constantly put in the manual effort to try to do that yourself. So this helps both with scale and with specificity, like both of the two problems that the classic methods had. So now we can be very specific, because you can have, you can dream up any topic that you might find online, and enter a few phrases to describe that topic. And the AI is able to take your meaning and find the relevant contexts that fit that topic. And it’s able to scale because it’s not just dependent on the words that you entered, but it’s dependent on all the related content that’s out there related to the context that you specified.
Vitaly (00:17:51)
Actually, I have a follow-up question for you. Obviously, when we talked about contextual targeting, we generally mean the words on the page. But I’ve gotten the question several times about whether we analyze images on a web pages to do contextual targeting. I don’t know to me, it sounds pretty difficult, because images can be so generic, but what are your thoughts on this?
Ned (00:18:13)
So there’s definitely the technology to do that exists. So there’s things that would take an image and sort of translate it into the content of the image, like this image has two smiling people on it, this image has a desk on it, and so forth. But that content is generally a lot less specific. I mean, think about images with like two smiling people. Is that really like a context? Is it sufficient for his brand? Yeah. Is it? Is it strong enough? So when you take a look at the, for example, the contextual technology that we have at StackAdapt, we make sure not only that there is context, but there’s sufficient context, that we can say that with a strong enough certainty that this matches what the advertiser meant. And an image might not be specific enough or strong enough to sort of relay that image. But the technology to do that definitely exists. And definitely I see in the future, a lot more contextual information being relayed with things like online video in streaming, CTV type content for to enable contextual advertising.
Vitaly (00:19:25)
So it’s more of a question of finding really strong intent signals and just with images, a lot of the times they just there’s a very loose connection with intent.
Ned (00:19:36)
That’s right. And the images oftentimes are not necessarily what determines the intent, though, of course, there might be cases where the images particularly strong, but those I think, are much less frequent case.
Yang (00:19:51)
I’d say there’s also two areas of innovation when it comes to contextual not just related to the accuracy and performance of it. One is the transparency of it, as well as it being able to auto-adjust and learn. So, you know, there are several contextual providers out there that have been out there for a while, and that claim to be improving their contextual algorithms. And that may be true, however, typically, it’s a black box of what type of pages to target, what type of pages they are targeting, and how well it’s performing. So the black box causes a very large problem, because when you input what you want to target, contextually, you’re not able to quickly double-check if this is actually accurate, or if it’s going to work well. So there’s a huge advantage of contextual being tied to the actual platform that is executing the media. In that case, the platform would also know and understand well, what’s the performance metrics of my campaign utilizing this contextual, then the system could auto-adjust to contextual algorithm. Order matches itself in order to make it perform even better in order to hit KPIs because at the end of the day, when you execute ad campaigns, your goal was to reach an objective. The input is just the initial learnings to help the system know, you know what type of contest you should look for. But in an ideal world, a true AI is a system where there’s a feedback loop, and it can continuously learn in order to ultimately just be hands-off and adjust itself to reach your objectives.
Matt (00:21:29)
Thanks so much. You know, I think on that note, this is actually a perfect point for us to take a quick break. So to all of our listeners, stay tuned for the second half of the episode when we do a deep dive on the future of contextual advertising and look at some key takeaways for your upcoming campaign strategy.
StackAdapt Ad (00:21:49)
You know, your advertising should be multi-channel, so why isn’t your ad platform StackAdapt offers multi-channel advertising solutions across native display video connected TV and audio with access to all of the major exchanges and more than 55,000 publishers as if that wasn’t enough. StackAdapt is also the highest rated programmatic ad platform according to G2. It’s time to elevate your digital ad campaigns. Request a demo at stackadapt.com. StackAdapt. How Agencies Thrive.
Matt (00:22:25)
Welcome back, everyone. Let’s kick off this half of the episode by looking at the role that contextual advertising plays in the future and more specifically, Ned fatale. Yang, I’d love to hear from the three of you about how you think contextual advertising should be a part of an effective media plan and why you think that this is important for those who are running campaigns in the programmatic space. We’ll start with you, Vitaly.
Vitaly (00:22:54)
Yeah, so to me, it’s all about performance. So if contextual advertising works for your brand, and it fits your business, then go for it. I think audience targeting is, is here to stay, I think it will change how it’s done. But truth be told is that there’s never going to be one strategy that fits fits every brand. So I think as long as marketers experiment in in a proper way to see how contextual could work alongside audience or what sort of results it can drive. I think they will be they will be in the in a good shape. You know, when it comes to planning, I always found that alo laughable when when people aren’t sure at least split up the media plan, you know, 15% through retargeting, 20%, that contextual, and actually execute all the way through with those numbers in mind. And a lot of that a lot of it is kind of pointing finger in the sky. To me, these should be just placeholders from where you adjust after depending on the performance of different tactics. And obviously, you can use also machine learning strategies to automatically allocate budgets based on performance of different line items. But I think contextual to me, is only viable if it if it actually works.
Yang (00:24:13)
Yes, Vitaly said each brand and situation is different, you know, contextual and user base behavioural targeting are not mutually exclusive. In order to maximize your potential, you most likely want to have both. Because for example, let’s say you only stick with user-based targeting over time, as more technology changes, as well as government legislation, the universe of people that you can target will shrink, because there’s going to be less consent and permission to target those users. So you’re not going to be able to scale your campaigns. And you’d be missing out if you were not using a contextual strategy that could also reach the performance that you’re looking for for a campaign. So In order to cover both worlds where you have your access to use your basic data and don’t have access to it, you would want both strategies side by side in order to really maximize it. And then you can monitor to see how well each is doing and then adjust accordingly.
Ned (00:25:15)
And another thing I would add to that is a little bit more to think about what a contextual strategy brings to your media plan that a behavioural-only plan doesn’t have. And basically contextual has an aspect to it that no behavioural strategy can match, which is it delivers your brand message when the user is in a particular frame of mind. Because your ad is appearing next to specific content that you’ve specified, you’re able to deliver your brand message when the user is thinking about something specifically. So for example, if you’re an athletic shoe company, you might want to deliver your shoe message when the person is reading about running online or something like that. And even so, that is vastly different than what a behavioural strategy offers where we would target runners, regardless of what domain they might be appearing on online, or what content they’re currently consuming. And this makes contextual, it gives contextual, the ability to be highly performant. And it brings something to your media plan that no behavioural strategy can bring.
Matt (00:26:29)
I do have a follow-up question for everyone as well, because I think at the beginning here, we talked a little bit about performance data. And I’m interested to know a bit more about what role performance data plays, especially during campaigns, and how does the future of this technology impact enforcing privacy for users?
Vitaly (00:26:52)
I think it’s at the end of the day, it’s up to every company to work with their legal counsel to make sure that they’re compliant with law. And I don’t think just doing contextual advertising alone is addressing all these questions related to privacy. I think it’s one of the pillars, but it’s certainly not one thing that you do, and then all of a sudden privacy consideration goes away because there’s a lot of things that happening. Also, on your own website, that are kind of independent from from contextual advertising. I would be hesitant to comment on how using contextual. It translates into the enforceability of privacy laws.
Matt (00:27:32)
In this last part, I do want to know from all of you talking a little bit more about the future, really, really looking ahead on this one about kind of what these future challenges look like for contextual. And how can advertiser expectations align with the performance of this technology.
Ned (00:27:52)
So, I can start there. So one thing that I say is over the past year, so StackAdapt, debuted our contextual AI tool in January of 2021. And since then, we’ve had a long time to collect real world data on this to see how it actually performs in practice. And we found that there’s a good fraction of campaigns where contextual campaigns either meet or beat the performance of some top behavioural strategies. So, for example, a top behavioural strategy that you might compare against is retargeting. And specifically site retargeting where the user has already been to the brand’s website and knows about the brand. And we’re looking for some conversion action afterwards. So it turns out that in about 23% of comparisons, that’s more than one in five comparisons that we had on our platform, and textual performed better than retargeting in these comparisons. So this is very strong evidence that contextual can be a real player in performance. But going back to in terms of aligning expectations, that doesn’t mean it’s going to knock it out of the park for every single advertiser. The context matters, the message matters, the product matters, all of these things matter. But the data has shown that it can be a very performance strategy. And it is really worth adding to the list of things to try when you’re going about exploring different performances for your brand.
Vitaly (00:29:29)
So I think one of the challenges for contextual advertising in the future is probably related to its brand. I think contextual advertising is still thought of as more of an awareness type of play. I think many marketers don’t realize that it can be driving results that Matt was just talking about. But aside from that, I think there’s probably a lot of things we still don’t know about the challenges through contextual advertising. We’ll have things perhaps, as more marketers adopt, there’ll be new market dynamics that may be translated into pricing or some new power dynamics that we’re not aware of yet. But as of right now, we’ve seen incredible adoption of contextual advertising, especially in certain segments of brands. So to me, for the next couple of years, it’s going to be it’s going definitely to become a staple for for any marketer. But how exactly will change? You know, over the last 10 years, in my career, I’ve been proven wrong so many times, so I’m hesitant to predict anything at this point.
Matt (00:30:36)
Yeah, absolutely. I, you know, Vitaly, I was about to actually just ask, if we think about 10 years into the future, what do you think things are going to look like, from a technology standpoint?
Vitaly (00:30:49)
You know, if we talk about advertising technology, if he asked me that question, maybe three years ago, or four years ago, I would maybe be in next camp with respect to AI, and how, you know, we’ll have engines that are so good at predicting people’s intent, and, you know, deliver messaging almost without any friction. But I think now, it’s becoming apparent that there’s a lot of resistance from users related to privacy with respect to processing data, there’s legislation related to privacy. So perhaps what we’ll see in 10 years, is going to all around, it’ll get more streamlined, more automated. But in the last couple of years, I’ve changed my mind about this, you know, AI that just makes decisions for everybody almost, to something more closer to what we actually have now. And still, maybe walking the line between, you know, user privacy, automation. And I think in 10 years, it’s going to be a lot more advanced than what we have now. But I think, probably not much more advanced. I’m not sure what you, Ned, and Yang think. You’re the technologist.
Ned (00:32:12)
So in terms of technology, I agree with you Vitaly, that there are some things that I would have a very difficult time replacing. So for example, when it’s your brand, and you have only this budget to spend for your first advertising campaign, there’s no amount of AI that’s going to substitute the kind of personal intuition that a marketer has, in terms of what the message should be terms of who they want that message to be directed to, and so forth. So there’s a lot of one time decisions, where there’s maybe not a lot of historical data on, it’s very difficult to collect historical data on it. And maybe the historical data is not relevant. And maybe there’s not a sufficient budget to sort of try many things at once and do a data driven optimization approach. So these kinds of decisions, I think, will never go away from the hands of people. But at the same time, I do feel that especially in programmatic advertising, the amount of data is just huge. There’s so many websites out there so many opportunities on which website which user to place the advertisement, where on the website, the advertisement should go, that a human decision maker just simply can’t handle all of those decisions, there’s too many of them for a human being to do. And that’s why I believe that there will be some mixture of these human decisions that are based on insight based on gut feelings based on directions and messaging, and AI-driven decisions that help us optimize all of the things that a human being can’t actively do throughout the day, by optimizing all the minutiae of delivering programmatic advertising campaigns. At the same time, going back to the original question of how I feel, advertising or contextual being in 10 years, one of the things that I feel is going to happen is when you look at the possibilities of where to advertise, as user data becomes scarcer, because of legislative moves because of privacy moves based on consumers and so forth. I believe that contextual information should walk in there to fill the void, we should have more information about what is the actual content so that we can deliver more accurate advertisement based on the content as opposed to based on the user because ultimately, the publishers are not going to be able to sustain their content without the kinds of revenue that advertising brings in. And given the absence of user data. The only thing that is left is contextual data on which to make the decision.
Yang (00:34:51)
If you see how a campaign is executed today, there’s still a lot of manual effort, you know, they’re still, you know, clients still like to optimize manually make changes manually, I would like to see that all go away, you know, even before 10 years, and have the human focus more on the creative aspects, the strategic aspects, even when it comes to creative, there’s a lot of manual testing and optimizations that can be automated. But I think as time moves on, the human will be needed and more of a high-level scenario. And a lot of the auto manual execution should be able to be automated in especially anything that involves data, or that involves testing things. Computers are powerful enough to do and learn all that on their own right now, it just takes, you know, effort, and a lot of precision for someone to design this accurately, such that it can perform to the level that you know, we expect it to perform.
Matt (00:35:56)
Now, I do have a kind of a practical question to ask all of you as we bookend this episode. And that’s more about the role that this technology and contextual advertising plays in helping out emerging areas in the programmatic space. So looking at something like connected TV, or gaming, or even kind of my favourite example, when talking about contextual is how advertisers navigate sensitive verticals. So I’m interested to know from the three you what can be said about these emerging areas, and how the technology is really helping each of these areas thrive.
Vitaly (00:36:34)
I’m going to try to take a stab at this one. You know, to me when I think about these verticals. And granted, we don’t do contextual advertising with CTV or I think it’s important to, to think of what contextual advertising actually means in those environments. So if you take gaming, for example, gaming, you know, there are some some big broad categories of gaming, for example, casual gaming, or, you know, or action games or RPGs. What does contextual advertising actually mean there? Is that the gameplay is it? Is that the category of the game? How does that actually fit all advertisers? And I suspect that, you know, if you look at gaming advertising advertising, right now, it’s dominated by other games. So I’m, I’m struggling to, to make a connection of how, for example, gaming contextually, would satisfy requirements for full advertisers in any category, there’s obviously, you know, these VR Worlds, and we’re in where you can have these virtual billboards. But obviously, they’re I question the scalability of these initiatives, because to me, you know, I don’t own VR headset, I don’t know how many people own one or regularly use it. So it sounds exciting, but I haven’t really crossed that, you know, mental mental barrier, over envisioning of how exactly will work in a scalable way.
Yang (00:38:08)
So the possibility of contextual advertising or even any other type of specific targeting for these different verticals or channels, comes down to what data is available out there and supported. So, for example, connected TV, contextual advertising is actually possible in certain cases, if you imagine and connected TV episode that, you know, is on a schedule, there’s actually transcriptions out there of the actual, you know, words of that episode. And that can be targeted contextually, certain providers support this already. But it’s a matter of having, you know, more access to this more support for this in the ecosystem. So it’s all a function of scale. So the more partners and providers can support, you know, transcribed connected TV episodes, then the more possibility platforms will be able to support in terms of more sophisticated targeting such as contextual on these channels.
Vitaly (00:39:15)
Again, would that ad be delivered during the ad slot? Or would that ad be delivered sort of like an in-screen while the video is playing?
Yang (00:39:26)
When you use connected TV, there’s going to be breaks for advertisements, kind of like when you watch a regular TV? So it’s during the episode, yes.
Vitaly (00:39:37)
Yeah. So that’s where I think I personally struggle to envision how it would work because you know, taking obviously, there’s some context, some some content will be pretty easy to target, right? If it’s Top Gear, for example, about cars, but if it’s, for example, a lot of TV shows, for example, Friends or The Office. What sort of contextual advertising information? Could you pull from that to determine the ad that you want to deliver? Right? So I think at that point, you kind of have to fall back on audience targeting.
Yang (00:40:12)
Yeah. Or you could try to locate specific brands or types of products being mentioned, in the dialogue, you know, of those shows essentially. So yeah, there’s a lot of different ways you can utilize the data, depending on what’s available, essentially.
Vitaly (00:40:32)
Yeah, so, potentially, that’s something we’ll see in 10 years.
Matt (00:40:34)
Vitaly, you perfectly transitioned to sort of this, this ending question that I had for the three of you, it doesn’t necessarily have to be related to contextual, but, you know, imagine possibilities are endless. Any technology can come out what’s that’s something that each of you think should exist in the advertising industry, right now, that doesn’t, sort of this wish list item that you’d like to see?
Vitaly (00:41:02)
I won’t reveal too much. But this topic is really exciting. To me, it’s actually related to creative. So there’s obviously a lot of talk about, you know, for years about how creative and media come together, maybe at an agency level or on the platform level, and how they work together. But actually, I think there’s a huge gap and a huge opportunity related to both of those. So I won’t reveal anything beyond that. But it’s related to creative, I think that has some of the biggest opportunities. So stay tuned.
Matt (00:41:36)
Ned, Yang, anything that you wish existed that hasn’t quite come out yet?
Yang (00:41:42)
Well, for me, I wouldn’t say it’s one thing. But in general, in the advertising world, it’s a very fragmented marketplace, you know, you have different partners, and data providers doing their own thing. So to achieve scalability in any one, you know, technique can be difficult. So, ideally, you know, I would like to see more universal standards, when it comes to working with partners, whether it be data partners, whether it be, you know, by side sell side, that would make it a lot easier and faster for the whole ecosystem to innovate and to integrate together. And to leverage each other in terms of, you know, all the different pieces, because, you know, in advertising, it’s very difficult for one player to do everything. If you do that, then you just become a walled garden center, and you’re doing your own thing. But obviously, there’s so many of these out there now, such that, you know, it’s really hard to scale across the whole internet. Because everything everyone is, you know, more and more. There isn’t a universal standard. And so, in order to innovate, you have to work across so many different partners in the ecosystem today.
Ned (00:43:00)
Yeah. And to echo a little bit of what Yang said, what I would really like to see is a more global approach to privacy. So we have a lot of regional approaches, with various countries and continents taking action. In terms of privacy legislation, we have various technology platforms taking action in terms of enforcing privacy. But this makes it very difficult for advertisers and advertising platforms to actually do our business. I think everybody in the industry recognizes that. Privacy is a really important thing for users. And everybody agrees that we want to deliver that. And so we’d like to see a little bit more consensus around the technology, and the methods that we can use to deliver that privacy, while at the same time being able to still advertise the products and reach the consumers that the brands like to reach.
Matt (00:43:52)
Absolutely. Well. You know what, thanks so much to all three of you for joining. This has been a fantastic episode. And I can’t think of a better way to have started the new year than with this topic and this roundtable discussion. So I’m sure all the listeners who have tuned into this episode have taken away as much as I have. And this has been the How Agencies Thrive podcast and I want to thank everybody for listening. Thanks, everybody.
Outro (00:44:21)
Thank you so much for tuning in. This has been the How Agencies Thrive podcast. If you liked what you heard, then there’s three things that you can do to support the show. Number one, subscribe. Number two, leave us a review. And number three, share our podcast on social media or with anyone who might find value in this content. If you have questions or feedback, we’d love to learn how agencies or brands work with StackAdapt, find us at www.stackadapt.com. Thanks for listening, and I’ll see you next time.
