At this years’ IBC (International Broadcasting Convention) in Amsterdam, DeviceAtlas’ Director of Product Strategy, took part in the panel discussion ‘Making advertising more viewer-centric’.
The discussion focuses on how platforms can evolve their advertising strategies to deliver viewer-first ad experiences that benefit both viewers and publishers. The speakers, John Leonard, Director of Product Strategy at DeviceAtlas, Igor Oreper, Chief Architect and Strategy Officer at Bitmovin, and Peter Verster, Founder and CEO at Northell, represented different parts of the advertising value chain. The talk was moderated by Ophélie Boucaud, a Principal Analyst at Dataxis.
You can watch the recording of the panel discussion here, courtesy of IBC.
Discussion Summary
John Leonard explains that DeviceAtlas provides intelligence on end-user devices, helping content providers adapt and optimize content based on device capabilities and limitations, thereby understanding user context without personal data.
Igor Oreper from Bitmovin, highlights their tools for video streaming services, including video compression, content processing for ad insertion, and playback across various devices. He emphasizes ensuring a seamless experience for viewers, including ad quality.
Peter Verster, CEO of Northell, discusses their focus on marketing activation, brand compliance, and using AI to understand video content better for metadata, tagging, and brand safety. He states that ads need to be contextually relevant and add value rather than detract from the viewing experience.
All speakers agree that viewer-centric advertising means enhancing the viewer's experience. This includes personalizing ads, ensuring smooth transitions between content and ads, and using device intelligence to infer user preferences and disposable income. They noted the frustration caused by irrelevant or repetitive ads and the importance of contextual ad placement to avoid disrupting the viewer's mood.
The panel also discussed the fragmentation in the advertising and content distribution value chains and the need for standardized data and transparency to build trust among publishers, advertisers, and viewers. Innovations like server-guided ad insertion and AI scene analysis are crucial for better ad placement and contextual relevance. Finally, the potential of the "second screen" (mobile devices used while watching TV) for interactive and actionable advertising was highlighted as a significant opportunity for advertising.
Full Transcript
O: All right. Thank you everyone for being here this morning. I hope you all got coffee, you're energized, you're going to follow the discussion through. We'll take questions at the end if you want to stay throughout the whole session. And I have the pleasure to welcome three speakers with me on stage: John, Igor, and Peter. They're going to introduce themselves right after. The session is going to be about making advertising more viewer-centric. More broadly, we'll also explore how to improve the ad experience for audiences in general. We'll look at the tech aspects of it. We'll look at the experience challenges. We'll look at innovations in this space that are interesting for the whole industry to be aware of and to explore. And without further ado, John, if you want to introduce yourself and talk a little bit about DeviceAtlas.
J: Sure, thanks. So, John Leonard. I look after product management at DeviceAtlas. DeviceAtlas is a company that most people won't have heard of, but we're very much in the background, under the stack, providing intelligence on end-user devices. As you know, if you're in the digital content business, whatever you're doing is consumed on a connected device of some kind. And what we do for our customers is we tell them precisely what device is being used at that moment in time by the consumer: its capabilities, what it is, and its limitations, so that the content provider can adapt and optimize according to the user context. So really, it's just ultimately about understanding contextual information about the user without any personal data at all. It's purely contextual information about how they're consuming the content.
O: Thank you, Igor.
I: Yes. Hi everyone. I'm Igor Oreper. I'm a Chief Architect at Bitmovin. Bitmovin is a software products company. We make tools for video streaming services. We don't make all the tools, but a few of the important ones that are relevant for today's discussion: everything from video compression, processing content to make sure ads get into the content in the right places, as well as playing back that content with those ads and doing it consistently and in a good experience across lots of devices. As John knows, there are lots of devices, and as we all know too, and also we make tools to help measure that experience for the service operators, everything from the quality of experience on the main content but also the ads themselves. You want to make sure that the viewer has a complete and seamless experience, and so our toolset is responsible for making that happen. Happy to be here and looking forward to the discussion.
O: Thank you, and Peter.
P: Hi, good morning. I'm Peter Vester. I'm the founder and CEO of Northell. We are a pioneering data and AI company. We were acquired by HH Global roughly 15-16 months ago. So the legacy is in marketing activation, so working with advertisers and brands, and we focus on activation content production. Our core focus is really on smarter, faster teams, and we focus quite heavily on unstructured data and making sense of that. As part of our products, we have Media Magic, which is effectively a platform that brings agentic AI into workflows, specifically where we focus on understanding video content better. Some of the core use cases are metadata understanding, tagging, but also compliance use cases for brand safety, regulatory compliance. And so we see that use case come up a lot. So effectively focusing on that as a core product roadmap and then focusing on data strategy and helping clients get value out of their data assets.
O: All right. So as you can see, we have people across the whole value chain of advertising at different spots. So we have the guys working with the advertisers and the brands, the guys working with the distributors, injecting the content and the ads all the way to the end viewers, and we also have the guys working directly on the devices that the end viewers are watching the content from. So I want to ask each of you what viewer-centric advertising means to you coming from the perspective that you represent in the value chain. Maybe Peter, you want to start?
P: Yeah, so I'll take two perspectives. So I'll take that of the advertisers. So it's around making sure brand consistency, brand compliance, the brand shows up well-formatted, well-structured, well-coordinated in all the markets. So we work with large companies who have a global local footprint. So that idea of going from a global brand to a local brand is really important to them. So that's primarily what we focus on. Then, as a consumer and a father of four, I think it has to add to the experience and not erode from it. So it still needs to sort of create value for me. No friction points in there. So I think that's value-centric. And then when you focus on, well, what does it actually mean? I think for the most part, ads need to be contextually relevant. So the wrong ad at the wrong time doesn't quite work, but a good ad at the wrong time is equally harmful. So these are the scenarios, the more nuanced, sophisticated scenarios is where we focus on. So I think centricity has a value creation aspect, and we want to minimize value erosion to think of it in those two sort of lenses for advertisers and for consumers.
O: Igor.
I: You know, from my perspective, viewer-centric really means it's about the experience for the viewer, right? So as an advertiser, you want them to see that ad. As a content provider, you want them to stay for the show and come back for the next one. And so it's really about how do you make that experience a really good one for the viewer? So, you know, for me, this is sort of analogous to how we run our business. At Bitmovin, it's really about the customer, right? And I think that's important to our end customers. So we build products that are really intended to make sure that experience happens. One example I'll make here is there's a lot of development in how do you personalize ads for viewers? And so a feature of a product we're working on now that's getting a lot of traction is server-guided ad insertion. And that's really about personalizing the ad experience and giving viewers choice so that they can select different types of display formats for ads, right? Some viewers may want to watch the main content right next to the ad because they're willing to watch the ad, but they don't want to stop the action. Others are willing to take a pause from the main content, transition to that ad, and come back, but you need to make sure it happens really smoothly. And so server-guided ad insertion is a feature, a new development that helps make that happen. And there are quite a lot of others like it, to your point on contextual advertising. You know, how do you do contextual advertising if you don't know what the content is, right? So one thing that we've been focusing on in the last eight months of developing is AI scene analysis. You've probably seen it at a lot of other stands here, and that's really about analyzing the content, multimodal analysis of content, audio and video, to determine what is the best placement of an ad, right? So that you don't interrupt that viewer experience and frustrate the viewer. What about, you know, what are the topics or themes in a particular scene that you may want to, you know, avoid certain types of ads or promote certain types of other ads? There's technology now available that makes that possible not just for new content but for your back catalog of content. And so a lot of our customers and prospects are really excited about this for the purpose of contextual advertising. And it's at the end, all about a viewer-centric experience.
O: So relevance of the ads, a smoother experience between content and advertising overall. John, what about you?
J: I think the context point is key, but what one could think about is also the relevance for the user. If you know the user is from a household, and you understand the household graph of different devices, if you're promoting a really high-end experience, there are some households, they may be aspirational, but they may not have the disposable income for that ad to really connect for them. So you can, just with information about the capabilities, the nature of the end device, you can infer a lot about disposable income, for example, for that household, and that allows you to choose which ad is most appropriate. A high-end experience or a high-end product might not be appropriate for this household; it might be appropriate for this one. So that's an area where you can tune the experience to some extent according to who the viewers are.
O: Yeah. Tell me from which device you watch, and I'll tell you what you're going to buy, basically, this idea. So John also suggested a very interesting question earlier about how we personally react to ads, because we're all consumers of content. So I just want to go around and ask each of you how do you react when you're watching a show and an ad pops up? Do you go away? Do you start scrolling your phone? Do you watch the ad very intensely? Maybe John, if you want to start.
J: Okay. I think there's a couple of factors in it. It really depends on the nature of the program being watched. So if it's a feature film which has ads, I'll probably wait through it because I intend to watch the next bit. If I'm essentially channel hopping with the family, I might just switch channels. Or another possibility, I might switch to my phone and scroll on Twitter until the ads go away, and then my focus will come back. But I'll probably keep an eye on it in case an interesting ad crops up that gets my attention. I may watch it. So I think it's a really, it depends question, but definitely there are displacement activities. I might go and make a cup of coffee.
O: Fair enough, Peter.
P: Um, personally, I avoid them. So I pay for my services. I don't want to see the ads. If I do something, it's by choice. It's largely informed by, I've got kids, so I want to make sure my kids see things that are appropriate for them, not influenced by that. And when I think about the economics, like YouTube Premium, 10, 15 bucks a month, that's what my intention is worth. But for me personally, the implications of not having that is far greater, but I have to manage my kids' behavioral differences. So I focus more on editorial content, educational content for my children as well. So I think I have a preference for not seeing things. I remember when there were only two or three ads on TV in the '70s in South Africa, and everybody talked about them because the production quality was high, but there wasn't a lot of choice. I think now it's finding the good ads amongst all the many things is a challenge, and so it's overload to some extent. So I'm trying to avoid that overload for myself personally and for the family.
O: Yeah. There you go.
I: I'm more with Peter. I try to avoid them when I can. So I, you know, pay for subscription services. More and more, there are hybrid subscription and ad-supported services, which I like personally because it's sort of fewer ads, or I get to make choices about when I want to see those ads. So if I'm sitting down to watch a movie, I'd rather watch two minutes of ads in the beginning and avoid them for the rest of the time. But also, I remember a time when, you know, I turn on Hulu, for example, and I would see the same ad over and over and over again, and that was incredibly frustrating. And I think we're past that. Thankfully, the ads are, there's more variety, they're more interesting, they don't interrupt the experience for me, so it's less of a barrier. And sometimes I'll get up and take a break when I, when I see that. So, yeah, it varies a little bit.
O: Okay. So yeah, you explored different types of frustration that we can feel. So the fact that we're forced into watching ads when there's no other tiers, no other options to consume the content. The fact that we also have a limited attention span and it drops significantly when ads are popping up in front of us. So we might have to do something else. What are the typical frustrations that the products that you're working on are aiming at, like done playing, basically? Maybe Igor, you want to start?
I: Sure, yeah. I mean, kind of along the lines of what I was just saying, ad placement is really important. And again, going back to AI scene analysis, I'm really excited about this because, you know, with all the Gen AI innovation that's happening, there's quite a lot that you can understand about the content in an automatic way. And, you know, effectively index all of the content library that you have to decide where are the best ad placement opportunities, not just the timing of them, which I think is the number one important thing, but also the duration, right? Maybe there is an interruption you can squeeze in for 15 seconds, but not for 60 seconds. And so being able to suggest the timing of an ad and also the duration of the ad is really important to kind of ease the frustrations and keep viewers around. That's one. And then two, again, contextual ads. So, you know, if you're watching a very intense scene, maybe it's a, you know, a dramatic scene that's very emotional, you know, maybe you don't want to put a super lighthearted ad in, or, or maybe a very boring ad in, right? It sort of disrupts the mood. So being able to match sentiment of the main content to the content of the ad itself is really important. And to do that, you know, using IAB content classification taxonomy, you know, which is more or less standard, you know, that said, I think there's a lot of standardization that needs to happen in the technology behind this, and I look forward to more developments there.
O: Yeah, and John, maybe we can go back to you. So the work on targeting also aims at making the ad more relevant to the viewers. How does that help with taking down those frustrations?
J: So yeah, a little to develop a little bit of what I mentioned earlier. Relevance to the user is really important, and you can pick up on things like brand sensitivity, brand loyalty. If a household has a set of premium brand devices, that sends a different signal than a household with emerging brand devices. Also, how old are those devices? Are they the latest model? Do they have two or three TVs in the household that are less than two years old, or are they all six years plus? And similarly, the specifications, do they have, are they viewing on a really high-end expensive device or really is it an older budget device? So you've got a huge amount of relevance straight away just from this. And yeah, you can use that to really increase the relevance of the ads.
O: And another thing that you also all mentioned was around the ad load. So if the ad load is going to be, you know, it's going to be five minutes, I guess the frustration peaks, and then you're zapping, or you're leaving the room, or just changing the content altogether. One way to take down this ad load and have more efficient advertising is just to make sure that the ad is relevant to the viewers, so you're not flushing irrelevant ads to the wrong people. How do you see it also impacting the way that streaming is doing advertising compared to traditional broadcast TV that didn't really have a choice in that matter? Maybe Igor, sorry, Peter.
P: Yeah, I think the benefit of what we work on specifically is taking a more advanced view on this, which is what we call psychographics. So demographics are fairly static metadata about programming, device, they're all static. So I could live in a household that's got old devices, but the moments, the mood in the household changes depending on what we do. So we focus more on that, and being able to respond to that is a far greater opportunity to actually get the right content in the right moments for the motivation of that psychographic profile, and then you can have temporal profiles as well. So maybe morning is different from afternoon, from evening. So you can be very precise with your targeting. So it's less about the static metadata, which are valuable signals, but it's combining that then with the psychographic profiles we can build up about programming content, but to do that, we really have to understand the content. We really have to be able to do that at scale to be able to then create those signals from the programming itself, layering on top of all the device profiling. So that's kind of more where we focus, which is a marketing thing. So if you think about propensity to buy, there are different propensities by different people in different times, and how do you slot into those are more important for us. So that's where we focus our sort of innovation stack.
O: So yeah, bringing together basically the contextual aspect and the targeting aspect. So you know what kind of households you're targeting, but also who is watching content behind it.
P: That's right. Yeah, that's right. That's right.
I: I would just add one more thing; history matters. So when you know about past watching patterns, you can also make some pretty good assumptions about what users will tolerate or are interested in. And so one of the things our customers use our analytics product for is past viewer engagement and ad engagement. If you know that a specific device ID, you get high ad abandonment rates at certain times of the day or days of the week or for certain types of content, then you may want to decrease, right, if you can control, and in streaming you usually can control compared to broadcast, the ad load. So you may want to sort of turn that dial down for that particular device or group of devices. So I think analytics and historical assessment really matters.
O: Yeah, it's interesting. So don't ever watch ads. So this way you watch even less because your device knows you don't like it. That's what I'm reading into. Okay. If we talk a bit about the technical aspect and the data foundations that's built on that, how do we know that the advertising value chain and the content distribution value chain also are very fragmented. How does that affect the way that information circulates across this whole value chain and how to improve that so that the data flows nicely and we can infer and build products that really have different inputs such as the device targeting and the contextual information on the content?
I: Yeah, so if I look at it from an advertiser perspective and see the sort of process by which campaigns get created, creative gets created and the work that goes into it, some of these things are in the making two, two and a half years. So there's a lot of dense information that actually does not make it to the devices or doesn't make it even to the sort of supply chain as it were in that end. So that all gets encoded into the products, but there's a density of information and fidelity that's completely lost. So even though you use demographics and all these other things to target this nuance within the content, and what we see is the content is actually encoded with information we need to be able to use AI to kind of reverse engineer as it were the content and the context out of it, applying it is really helpful and doing that at scale is quite important. So it means an advertiser has a specific intent encoded in the sort of creative, but that information around it gets lost, and how we can resurrect and make it part of the supply chain is really important for us, but fundamentally doing that at scale. So we see a lot of manual taxonomy coding, different taxonomies. Those challenges are all everywhere. But if you can map the content to an existing taxonomy, and there can be multiple taxonomies, freeing up that need to standardize is quite helpful because getting multiple organizations to standardize is almost an impossible task to move them all at the same time, let alone convincing them to move. So that's where we see the benefit of what we do. The scale of that almost contextual tagging of the content within a given context or multiple contexts at the same time is what we see the position.
O: And John, you're supplying data to a lot of different platforms. How do you also manage this fragmentation and make sure that the analytics you have are aligned across everything that people also understand the data points as what they are?
J: Sure. A key point is we supply quite a lot of analytics platforms. So the publishers that are in receipt of that, using those analytics providers, they have consistent data then, often with the SSP. So we work under SSPs, the publisher and analytics platforms, and that helps to unify the understanding of the device. But the programmatic supply chain, you've got multiple stages. You've got the ad server, the SSP, the DSP, and the SSP is responsible for passing that data downstream. So you've got the device object within the whole bid request, and the device object carries those signals about the end user device. So as long as the SSP is passing clean data down to the DSP, the advertisers, their campaigns will fire reliably. As long as you've got consistent data along the whole chain, then you're in good shape. The problem arises when you've got different sources of data, different quality levels of data. Then you've got the advertiser essentially, potentially with a different picture than the publisher, and then you get discrepancies. And that reduces the overall dependability of the whole chain, but ideally you want a consistent, clean source of data across the whole chain, and that helps a lot.
O: So what's your role in cleaning up all this data? What can you do to make sure that this remains smooth and to avoid?
J: We encourage our customers or businesses to use reliable sources. Obviously, there are many, many sources out there. There's open-source data. We amazingly still find organizations who are attempting to build their own datasets, which continually surprises us because it's really not trivial and it's really quite dull in reality. But a surprising number of businesses still think ‘I can do this, I can do this myself’.
O: And create another silo. Yes.
J: Yeah, so they end up building and maintaining a solution which will never be optimal. It'll never fit in with other data sources in the chain. So really that's the role we play. We try to provide really clean data where the provenance of the data is transparent all the way down the stack, and clean authentic data with a clear provenance. Ultimately it's about confidence and trust.
O: Yeah, I'll go back to that right after. Good point. And what about the content distribution side? So what can the publisher do to make sure that this whole chain has clean data all the way through?
I: Stop building new data silos, or if you do build new data silos, connect them to the existing ones, right? Um, yeah, I can only second what John has said. The benefits of streaming is, you know, you get all this personalization and distribution and I mean so many benefits, but you have new data coming up everywhere, right? And then you have folks who are building their own data sources and they're not connecting it to other existing data sources. They're not using taxonomies that are already defined and out there. So I would say if you're a vendor who builds technology, you know, adapt existing standards, existing taxonomies. If you're a publisher who is, you know, in many cases collecting a lot of this data and information, use existing tools and standards to do that. Don't invent new ones. If you're an advertiser, same thing. So I would just say stop inventing new stuff, right? We have plenty of existing standards and frameworks for data management, connect them. And I think, again, you know, bringing Gen AI into the picture, right? It's very dependent on data. And so if we can connect more of those data sources than, you know, agents and MCP services, they can really make much better use of it and make much better sense of it. So you know, one thing we do for example in our analytics product, because I realize that's you know our own little silo of data, is we adapt a standard. So we have, you know, 300 different data fields that are standardized, but then we give that data back to our customer, the service operators, in a programmatic way so that they can consume it and connect it to their other data sources, application data, advertising data, and that way it's portable and it's actually useful for them, you know, beyond it just outside of our tool. So it's a big problem, but also a huge opportunity.
O: That's a great conclusion to that point. And now going back to what John started building upon around the trust issue. When we have a fragmented industry, obviously we need to build trust, and it takes time around digital advertising. We've had a lot of scandals, a lot of breach of trust as well between publishers and advertisers, also between advertisers and viewers. How do we make sure that we can create a healthier environment for everyone? Maybe first starting with brand safety and compliance. Peter, go ahead.
P: So, that's our primary product when it comes to advertising is making sure the brand's compliant and there's regulatory frameworks that they need to be compliant with. So that's the primary use case. But as a result of that, what we can also then do is build in some of our additional products to signal contextual relevance. So like I mentioned before, you know, a good ad in a bad spot is still as damaging as a good ad, a bad ad in a bad spot. So that's kind of what we're trying to mitigate. So having that sort of relevance and the ability to cycle through new signals to then adapt on the fly as it were is going to become more essential over time because bad news travels faster than good news, and you only have to be associated with the bad news to have damage to your brand. And so making sure that you associate yourself with the right content is part of what we do. The scale is the problem. Understanding the content at a deeper level, which is what we focus on. So as part of the compliance process, we have a deeper understanding which becomes the basis for then contextual relevance, and then making sure we can apply that with additional signals that might not be in the ecosystem. So newsworthy items, for example, plane crashes might mean that you don't want to advertise flights. Those kind of things, we can really zone into those. So just being contextually relevant and moving quicker with a pace of information is what we focus on effectively to make sure that brand compliance is beyond just the mechanical brand compliance checks.
O: Yeah, it's also an example that pops up a lot when we talk about the nuances in context. You don't want to put an ad break with a new car brand within the car crash scene in the content. That feels bad. That's not a great moment. Yeah.
I: Yeah. Um, I mean, contextual, right? It's, uh, if you don't understand the context of the ad, then a lot of things can go wrong. Um, I don't know. I've heard that like any publicity is good publicity, but maybe that's not true.
O: Um, you can ask Sydney Sweeney maybe.
I: So from my perspective, uh, you know, our goal at Bitmovin is really to make a tool set available to our customers to, to make all of this possible, right? Um, to make the business logic and business decisions around contextual advertising possible. And so that's really what we strive to do with, uh, AI scene analysis, being able to really inspect that content, understand deeply, what is happening, and not just, you know, the sort of the people and the objects, in a particular scene, but what is the mood and the sentiment and are there sensitive topics? Um, and, you know, it's very easy to build a rule set that says avoid this type of advertising around these types of sensitive topics, or maybe promote other types of advertising instead. Or maybe just avoid the ads in those topics altogether. So, yeah, I think our view is, we really want to bring the best tooling possible for the streaming service operators to make this happen.
O: Yeah, it's also like a legal thing at some point when you don't want to push certain types of content to younger audiences, or you know, like no alcohol ads in some countries after a certain hour, things like that. So you absolutely, absolutely need this kind of tool. Another point with the trust piece is around the viewers. So of course when we work with a lot of data to understand who viewers are and how they will react to ads is the privacy issue. So since John was talking a lot about targeting, I'll start with you. How do we make sure that viewers are protected, viewers' data remains private and go sort of around that to still make relevant ads?
J: Well, our stance on that, we try to avoid any personal information. We, in every jurisdiction, more or less, information about the device they're using is not deemed personal. Um, so there are some contexts where even that may be viewed as not invasive, but kind of borderline. The more you know about somebody, you know, you can build up a picture. But generally, knowledge of the device is deemed kind of non-personal. So in a way, we kind of avoid that whole problem in effect.
O: Yeah, you're not looking at what people are searching online or you're not, exactly. Yeah. You're just looking at what's in, what's in the household gives a good picture, like a picture that is bright enough. Yeah. And Igor, I think you also mentioned another thing that was interesting at the beginning around the choice. So leaving choice to the users to also be affected by ads or not, and like having this choice sometimes just by subscribing to a paying tier instead of having the ad supported.
I: Yep. Yeah, I mean, you know, because we have the ability to interact with users, we don't necessarily need to know who they are and where they live and all of the details about them. But we can, we can ask them for what their preferences are, right? And if they tell us, then we can use those preferences to personalize the ad experience for them. And again, I don't need to know anything about them, but I know that if they're on a mobile device, it's probably an individual person. They have a unique device ID, but that's it. I don't need to know anything else about them as a person. And all I need to know is that they've told me, I wish to watch an ad load, a bunch of ads in the beginning of my content instead of throughout my content. And so now I can make that, I can honor that choice, right? And give that user what they want without knowing anything about who they are. And, you know, maybe over time, they want to share their viewing habits with me as a publisher, and help me make better decisions for them. And I, you know, you've probably all had similar experiences on some of the streaming services you consume. I'm seeing that more and more; user choice.
O: And going back to John, when you started talking about trust, it was also about building a trustful relationship between publishers, device owners, device manufacturers. How do you help with that picture?
J: I mean, I think at the bottom of the stack of trust, you've got transparency at the bottom, and that enables building confidence and ultimately trust. So if there isn't transparency, you have no chance of building trust. And in the AdTech chain, the transparency of data is really paramount. There was a big change in the industry three years ago when the main browser started withholding information from the HTTP headers, and ultimately that took away transparency from whole sectors of the AdTech community. It was still available, but you had to jump through hoops to get it. So there was a big reduction in transparency, which impacted different people in the chain differently, and that kind of reduction in transparency ultimately is damaging to trust between entities in the ecosystem. So really, we very much promote transparency because ultimately that builds confidence and then trust. And if everybody's working from the same data, it eliminates so many sources of friction.
O: Yeah. So making sure the ad is going to this device, and this device is a CTV, or this device is a tablet, and like not something in between or something blurry.
J: Yeah, we, it's quite interesting because we do occasionally get organizations contacting us saying, "Can you please recognize our device as CTV because we make more money?" And our commitment to the industry is we, we tell them exactly what the device is. But it's interesting the pressure that there can be from organizations that would do better if we misidentified their devices.
O: Yeah. Because of course, the pricing is not the same.
J: Exactly.
O: Yeah. Um, so we only have a few, like 10 minutes left. So I wanted to build up on the innovation side of the industry. What are the innovations? It can be formats, can be targeting opportunities, it can be measurement tools that you're looking into that you think are really important for the industry to be aware of at the moment and to, yeah, go deeper, um, to be integrated, maybe Peter.
P: Yeah. So from our point of view, really simple, look at operations and see if we can find efficiencies within that. So we think about processes, content production, distribution, that whole cycle, and just what can we do in there to just make things more efficient, more effective. So at a base level, that's what we focus on with our processes, our tooling is around that, that stack. But when we think about what technology makes possible now, it's this idea of building psychographic, um, profiles in real time and responding to those. So shorter moments we can target rather than, uh, static, you know, metadata around devices, which are more long-lasting, still valuable signals, but don't change as often. So we focus more on the dynamic nature of profiling, if that makes sense. And then being able to do all that at scale and do it more accurately, more consistently, but within specific contexts because as you find global local brands, what we expose ourselves to quite a bit is there are very different nuances in each under the locals. They're not all the same. It's not a homogeneous ecosystem. So being able to be nuanced, specific, and sophisticated is really valuable to the advertisers because it actually helps them create more value, but also limit the erosion of value potentially. So that's what we focus our use cases on. And I think all of what we do is focus on that primary use case and all the R&D and spend we make is in that space to make that more operationally efficient, but also more applicable, more relevant in shorter sort of time frames, that makes sense.
O: Like real-time analytics it is also very easy to adapt the campaigns very fast, react.
P: Yeah.
O: And that's very important in the performance-driven era of marketing.
P: Yeah.
O: Igor.
I: Yeah. So one innovation I mentioned earlier, server-guided ad insertion. Again, really a lot of interest there. It's really a hybrid of client-side and server-side ad insertion. The problem with client-side is ad blockers. The problem with server-side is scale. And the problem with both of those is lack of personalization. And so server-guided really intends to address that. It gives the user choice about how they view the ads, and what the display formats are. It allows distributors to scale their services to massive concurrent user numbers. So, when you think big live sports events like 300,000, 500,000, or a million-plus users, server-side is really hard. In fact, most will just avoid those ads altogether, and there's no dynamic ad insertion for those really big live events because when you fail at a big live event, everything goes wrong. So the risk is just not worth it, and server-guided addresses that problem as well. So, we really focus on innovation there. And then on the measurement side, there's not a lot of data for the streaming service operators across platforms, or rather, the data is inconsistent. In fact, if my event is running on a certain platform or device, I may not have any visibility into what's happening with those ads. And so it's really hard for me to validate what revenue I'm going to get. I just have to trust the system, and so there's a lot of innovation we're doing on the measurement of ad insertion, you know, the engagement rate with ads, completion rates, abandonment rates, but also the quality of ads. And, you know, are the transitions happening well from main content to ad and back to main content? And that really gives validation, trust, and confidence across the value chain because there's real empirical data that says, "Look, here's what happened and here was the experience for a given ad during an event." So I'm really excited about those two.
O: All right, so server-guided ad insertion, Bitmovin has a blog post on that if you want to know more that you shared with me so I could understand the concept better, and better validation metrics. So innovation in that space as well. John, how could we utilize device intelligence to also build upon innovations around this space?
J: I think an interesting thing with ads is the second screen is a massive opportunity in AdTech. So you've got the main content on your TV. You can link a mobile device to the same session just by scanning a QR code and you've got a linked device. On the second screen, you could show a contextual ad. So on the main screen you've got a dinner. On the second screen, you could pop up the recipe. You could drill into more information. You could order the ingredients and they'll be delivered next day with the recipe. So you can create product placement out of almost any content on the main screen by linking the second screen, it becomes actionable.
O:Yeah.
J: And there's so much variety, or there's so many things you can play with based on the main content that you can surface on the second screen. And in a way, the second screen is an intimate, close device. Everything becomes actionable. So I think a real opportunity to explore is that linkage, the real-time linkage, and take advantage of the second screen.
O: Yeah, it's also a good purchase funnel. It's also something that a sports streamer can leverage for showing stats and interacting with different pieces of content. That's a very, very strong pinpoint that we still need to unlock. I think I'm not sure that many streamers or content providers figured out how to really create a bridge between those two.
J: It's early days, but the beauty of the web or addressable TV is you can create a session token that links the two. So just some connection between the device, and scanning a QR code on the screen is one way to do it. There are others. But it's very easy and it enriches the user experience.
O: Yeah, it will. Okay, final question: How would you advise people in the industry to go towards more viewer-centric advertising experiences? What is the one thing that we still need to unlock, and if you have one key message to share with the audience?
P: I think it's easy to assume that CTV is the entire advertising ecosystem, and from an advertiser's point of view, it's not. It's one of the channels. So if I was to have a magic wand, it would be able to track effectively and measure across all the touchpoints of a user journey because that's actually what matters, because you engage with the same brand in multiple touchpoints, CTV being one of them. I think that would unlock a ton of value, and it sort of builds on your point about actionable, but I think it's almost a step before actionable. It's making sure you show up consistently at the right time in the right moment across channels, touchpoints, and in our world that includes physical, which is interesting because people tend to discount that, but it's a huge part of still exposure. So I think that would be my magic wand moment. If we can make that work, advertisers would be far more effective in actually creating content.
O: Igor.
I: Yeah, I think make use of the technology available to us. I mean, there's so much innovation that's happening in this space, and don't forget about the end user. The viewer is really at the end of the day what makes this all work. So if we as an industry, you know, help them, then I think all of the metrics go in the right direction, and we have the tech to make that happen. So let's make good use of it.
J: I would add to what Igor says. The technology is available to do some pretty amazing stuff. And one really simple innovation could be if your business is about subscription content, which Peter was mentioning earlier, subscription sharing is generally seen as a big problem. But actually, if you're monitoring, you have visibility for everybody who logs in, what devices they're using. So you might allow John Leonard to have three or four different devices that he can consume content on. But instead of drawing a hard line and blocking other devices, instead of that, just monetize through advertising. Use ads for any subsequent ones. And if they're legitimate, you can just give them an extra security hoop to jump through. But other people who are essentially freeloading on the same content, you're monetizing them.
O: Yeah.
J: Instead of having the PR problem of turning them away.
O: Yeah, exactly. That would have created a different approach for Netflix, for example, when they switched from the big sharing account policy to no sharing account at all at the same time that they were launching ads actually. So yeah, that's an interesting point as well.