Field Garthwaite, Co-Founder, and CEO of IRIS.TV, joins Michael Beach to discuss the power of contextual targeting, and how publishers with high value inventory can use video-level data to improve performance and justify higher CPMs. Watch our latest Screen Wars Thought Leader Interview here and read the full transcript below!
MB: All right, Field, welcome to Screen Wars.
FG: It’s great to see you.
MB: I’m excited about this talk. Give us the overall story with IRIS.TV and what problem you solve.
FG: IRIS.TV is a unique video data platform. Our core technology is a content identifier for streaming, very similar to a webpage URL. Its name is the IRIS ID. In short, IRIS.TV is making video data accessible to buyers and sellers. It also impacts consumers by driving better viewing experiences, thanks to having better fidelity data on what’s watched. We have a recommendation engine, and we also have a lot of products being built off the IRIS ID. These products are things like contextual targeting or measurement, that help you understand what the viewer is watching, so that you can make sure the right advertising aligns with the right content.
MB: Who is your ideal customer?
FG: We have two types of customers. We work directly with content owners, broadcasters, streamers, and big streaming platforms, like Vizio or LG. These streaming platforms are becoming the new cable companies, because they put everything together on the screen. We also work with data companies. We act as middleware, as this infrastructure player that helps connect the video of these content owners to all of these different data products for contextual, emotional data. This includes things like logo recognition and brand suitability at that exact moment. So you can actually understand the data.
Leading data companies, like Oracle, GumGum, Pixability, and IES, are investing heavily on computer vision.
They break down the videos and then create standardized segments. So you get the same data segment across thousands of publishers and streams, and tens of billions of available ad inventory of people watching a show at that moment. For example, if you’re 1-800-FLOWERS, you can actually reach people watching a specific show right before Mother’s Day at scale. You can reach a billion viewers at that exact moment, when it is relevant to them. We act as that infrastructure between the content owners and the data companies to make all of that signal available on the ad platforms.
MB: If I’m a buyer trying to utilize your product, how does the process look like? Is it something I do it through a DSP?
FG: It depends on the way that the buyer is activating. We have integrations with ad servers, so the publisher can activate it through their ad server. That allows the content owner to package and sell their inventory. For example, let’s say you are Kohl’s. You want to reach consumers at scale when they’re watching baseball, or watching cooking content, since that’s an incredible time for the messaging and the type of sale that you have going on in the store. You can do that, and you can do that directly with a content owner.
But we also work with the SSPs to package Private Marketplace Deals (PMP deals). You can reach consumers when they’re watching cooking content across multiple publishers at once. We do that with all the leading platforms out there, from Magnetite to Xandr. Then, you can buy it through DSPs. We just started rolling out direct DSP integrations with the IRIS ID to enable the data partners that we have built on top of it. So products like contextual targeting can now be available in the DSP, but to date, most of the activation is happening through PMPs. We’re expecting that to continue because that’s a huge channel for connected TV buying, but it’s all three. Anywhere you buy your CTV ads, you can access this data as long as there’s an IRIS ID on the bid request.
MB: That’s interesting. We have people doing data-driven linear and things like that, and you can get info like a rating, and a content library, and you can figure out that there was a specific show on at 8:00 PM on TBS on Monday night. Streaming doesn’t have that data available, and it seems like not knowing that breakdown causes a ton of friction among many people we work with. Is that what you’re seeing in the market?
FG: Yes, we hear that from buyers all the time. The common buyer expectation is to put a show on television and have the exact same data in CTV. That often comes from people newer to that environment. And CTV is new, it has not been around for decades like television, so there are some complications that come into play.
A TV buyer will frequently buy based on identity, or they’ll buy a programmer that they really want to purchase inventory from, and they understand where they ran. They could also do some type of package, where they’re purchasing certain types of inventory as a bulk buy.
The content owner has it very easy to manage all the information they get on where that content ran. There are a hundred thousand slates in terms of places that they could run that year. It’s not that much scale in terms of the actual avails and units that are available compared to streaming.
Streaming is closer to a digital medium, since every individual household is seen as a unique stream of data. It’s on demand. They’re watching the content when they’re tuning into a VOD, or they’re tuning into a fast channel when they turn their TV on, or they’re scrolling, or they have a subscription service that offers advertising. Both the content and the ad placement are unique to that user and that household. There’s a lot more complexity.
And going back to that the example you shared about the show episode, doing that is now illegal due to the Video Privacy Protection Act, that prohibits sharing personally identifiable information. It considers data like your address, email, and even your IP address as personally identifiable.
And the other side of it is the program information. So sharing those two together is a big problem. In this industry, those are often things that you’d expect to be coming together from the digital side. There’s always a URL and targeting based on identity and IP address and IDs.
Streaming has a unique challenge since this data is not available and is now illegal. It’s a very hot topic, and there are so many class action lawsuits going on right now. Hashing that data is the best way to limit liability as a content owner. It provides buyers the transparency they’re looking for, and the ability to use the actual content and contextual nature to target.
The way you can hash that data is using a content ID, and this is one of the major reasons why we created the IRIS ID. Now you can have all of this information stored in one identifier. Technologically, it is really similar to a URL. It’s passed through on the ad call, but the ID stores all the information about that piece of content. That would be the contextual segments, the emotional segments, logos, and things of that nature. While the ID stores all that, we don’t make it available unless it’s actually contracted. We hold that information very close to the chest on behalf of the content owners. Then, we ensure it is available in environments that help them generate more revenue, and that it is not just something that is made available and then generates more issues. We’re very careful about how we manage data because generating more revenue for content owners is what we do.
MB: Continuing with the content owner example, what’s an ideal use case for a content owner working with you?
FG: The main use cases to date have been programmatic, and they’re starting to turn over into direct use cases. For example, a major auto brand or a big tech company use contextual segments to make sure that their messaging aligns with moments that are relevant to the brand, overall driving awareness. The most common way that we see publishers opt into this is by participating in private marketplace deals that are being purchased at large scale. And increasingly, we’re seeing publishers package it up directly and offer it. The inventory on LG will be very different from the inventory on AMC networks. They have different types of content getting viewed.
The packages of segments that AMC networks offers compared to LG or Vizio, are going to be different, and they are going to base them on where they have the most scale. That’s a technique that a lot of sellers are considering as a way to help them increase the reach of a campaign.
The biggest use case right now is when someone is trying to buy niche audience segments, and they already have certain audience segments. If they’re open to align themselves contextually with content with a lot of available inventory, that helps the sales teams to make sure that they can fulfill campaigns. That way, that sale doesn’t go to another company, like YouTube. And we’re excited, soon we will announce a lot of new capabilities around the data that we are introducing in close collaboration with the content owners.
MB: In our newsletter, we write a lot about Wall Street’s expectations on the video ad market. They’ve got the TV market growing about 4% a year out in the late 2020s. But as the share of time goes from linear to streaming, the ad load is dropping so much that we’re losing about 8% of the impressions a year. And if people are not watching TV as much as before, the revenue is going up, but the impressions are going down, then CPMs have to be much higher in streaming than linear for that to work. And I get a lot of pushbacks on that from people. They say they can’t get buyers to budge. They do these package deals, and they charge the same for the linear and the streaming, even though it’s got 75% less of an ad load.
Do you see content owners that utilize the IRIS ID driving up CPMs? And if so, give us an idea on that.
FG: There are two examples I bring up in terms of how to increase CPMs. The primary ways that have been using to help increase CPMs comes down to improving performance, the emotional resonance, and the actual impact of a campaign. So let’s talk about that first.
The ability to use what we call “Video level data,” actually understanding what’s happening in that TV show, movie, or video clip, at the moment when the ad is running is a new data set. You can actually tap into the moment and the exact frame of mind that the consumer is in. Think about when are we watching TV? It’s when we’re relaxing, we’re with a loved one, or whatever it is. We’re in a mindset that is very specific to that moment, and we’re spending time with the content. It’s very different from scrolling. It’s very different from being on a webpage.
This is why that inventory is so valuable. It is about the ability to tap into mindset, which is now possible. You have the big screen, and now with video level data you have these capabilities to tap into that mindset. It’s been proven that if you can align ad creative with the content that consumers are viewing when they are in that particular mindset, it drives performance.
Magna and IPG put out some research on this around the performance of mindset. And they had some surprising findings that counter a lot of what we think about buying digital advertising and the importance of identity. Some of those things were the increase in search intent, things like brand recall, unaided ad recall, and even sales. That was an amazing study that they did with a terrific product from GumGum called Verity, that does unbelievable computer vision analysis.
And there’s more research coming out about this. Things like, the impact that happens if you don’t have that type of alignment. There are things like negative sentiment, which is actually worse than running the ad at all. Those are things that are few and far between. But the point is that different messages from different creatives and advertisers are going to be a fit at a different moment. It might even be a sad moment in a movie, but there’s actually joy as well at that moment. There is relevant advertising that can appeal to that. Advertising that hits that same emotional cord actually drives recall, and favorability, and awareness. The viewer is so attuned to that moment in the content that if the advertising matches that, you actually have the highest quality attention.
So that is driving performance, and it’s driving the outcomes, and the things that people want from TV advertising. This goes back to that old adage, “I know 50% of my advertising is being wasted, but I don’t know which 50%.” We’re helping to actually know where to spend more. And those segments aren’t necessarily always the most available. For example, the content around college basketball that Paramount is selling. There are also a lot of clips, and you can target those types of moments in and around the tournament. That’s inventory that’s going to be very high in demand, and it’s going to be at a high price point during that period.
The same goes with other things that are high in demand, just certain identity segments. When there is a low available supply, the price goes up. So we’re very focused on doing more research in that area with our partners, publishers, data partners, ad platforms to just put out the value in the ROI. We’re excited that we’re cooking up a few things right now that are really meaningful.
The second area is that there has been growing concerns in CTV from advertisers that don’t know where their ads are running. Some research has talked about liquor ads running in kids shows, or political ads running in cartoons. Things that historically have been illegal in television. This new format has grown so quickly, and there are so many avails that are all programmatically targeted, that it’s hard to control.
So, for the suitability side of that, we found that tons of news inventory is getting filtered out. And it’s similar with content that has high risk of guns. Those are pretty small select moments in an overall piece of content. And news in particular are vilified and blocked.
We’ve done research and the Oracle Contextual Intelligence Unit has published research that shows that 97% of the news is safe content. Among all the content that consumers watch to consume information, news has proven to be one of the best mediums to drive sales, and most of it is safe.
So by being able to identify the small percentage of inventory that’s not safe, and that is concerning to a particular brand, that inventory can get filtered out. The rest of the inventory is super high value and is helping to dramatically increase revenue for some news publishers.
Those are the two areas that I had mentioned. And the last thing on that is the premium price point that’s paid to content owners that are all the broadcasters that we know. Now, some of these new OEMs platforms like Roku deserve those premium price points. They have some of the best content in the world, and they are able to align it with data. And you know that you’re getting a quality experience. They’re now introducing video level data and a level of transparency that’s unprecedented.
It’s in stark contrast to YouTube, which is just trying to drive the lowest possible CPM and trying to sell through on tonnage. You have guarantees that you’re along brand suitable content, but we just found out from the Wall Street Journal’s research that those guarantees are pretty hollow.
They make you opt out at the last minute from buying the Google network. There’s no content that the ads are running against, it’s just playing on pages. The sound is off, and it’s going back three years. For example, a movie studio. If you buy a lot of impressions on YouTube, those impressions that didn’t have sound on, when running on network sites, counted towards your overall reach and frequency measurement against a household and a user. It takes five to ten impressions to get someone in a movie theater. But now you don’t know how much of that was wasted.
In general, there should be a stronger push to pull back spend from these opaque environments that don’t leverage things like video level data. Because you have no idea where your ad is running, or if it’s even alongside content. That’s a strong contrast from all the companies that we work with.
MB: Yeah. I thought that the WSJ research was crazy for a couple of reasons. We had Mike Shields on some months ago, and he wrote a piece about the YouTube ad market flattening out. We talked a lot about it and the complexity created on the buy side, where they made it really hard to not include that inventory.
And if you wanted a certain type of target measurement, then you have to opt it in. But even then, digging into the measurement partners, how did all the measurement partners not catch this? And the way they set it up… AdExchanger went deep on it, where they basically prohibit the measurement part of even having enough data to find out that was happening.
FG: It’s the same old story when it’s mediated by the platform. I think what AdExchanger called out is that there’s no pixeling happening, which is how it’s done on the web. There’s not an IRIS ID on every single YouTube video. There should be, but there’s not. Without that level of transparency, and if it’s managed only by the platform, you’re just trusting the platforms, and then you end up with a subpar product.
MB: Back to the big picture question about the market. I talk to people, and I always bring up that the CPMs have got to go up by X percent just to keep the market flat. And I often get feedback saying, “We just can’t get the buyer to increase. We’re competing with a CPM like YouTube.” Because people wonder, why does it matter so much that they did this? And I say that they’re actually harming the market. Because a really high quality producer can’t go in and get $40 or $50 a CPM. They should, because of the tech and the content and all those things, because they’re competing with a $20 or $15 CPM over here.
FG: Or much, much lower.
MB: Much lower, but that is actually not real. But they’ve made it almost impossible for people to figure it out. And it has a big impact on the market. I think people miss that for sure.
FG: That’s why I think it’s important to call it out and to call out when people are trying to sell make belief. Because it’s not the same thing. It’s really important that everyone knows it’s not the same thing.
People create models to understand the amount of impressions and time that someone needs to understand my product. How many TV ads they need to see to have awareness to then drive a sale, or drive a lease of a car. And if that is all out the window, and you have no idea if they even saw it, or if the sound was on, then how do you bring the science that there’s been so much investment to? We have to question every MMM model at that point.
It just goes back to your point. It’s something that really messes with the market. It’s something that the premium side takes the hit, and they’re everyone’s target right now. Even though they are the entertainment companies that we depend on for all the shows we love, and all the news we love. I think that they need to put less content on YouTube, and people on the buy side need to be smarter about knowing what they get on YouTube and that it’s not the same.
MB: Looking at the big picture, where do you want IRIS.TV to be in five years?
FG: We get really excited about the general technology capability, and video level data. There are many advances going on in AI right now that also impact the types of models that our data partners use. The types of models that our data partners use are getting better and better. And we are excited about the introduction of things like emotions, logos, and other types of data sets that are getting built around video level data.
Part of what we would like to see over the next five years is that people treat video in the same way that they do search. And even YouTube, and the way they’re buying the contextual nature and trying to own certain categories and channels, but going even more into the digital side. When you think about search, it is a medium that it’s always on. It’s something that you’re buying so that you can be directly aligned with either the key search queries, and other key elements that you know your brand will affiliate with.
In the same way, if you know what those are, you can buy those same topics, those emotions, those sentiments, those exact moments. That can be always on, and you can make sure that you’re affiliated with those anytime a consumer is sitting down in the right mindset for video. That’s really where we would like to see this go. And the reason that is so important is because that unlocks the performance, and the search and social advertisers investing more heavily in CTV. That’s a $180 billion market. There are around 9 million advertisers. We all know that the Fortune 1000 are going to continue to buy TV. That’s going to transition into video and streaming and CTV, but the performance side of the market is really important.
The other thing that we would like to see happen is the incorporation of our data into ad creative. We want to see that happening using the same computer vision AI to understand what emotion is my ad creative? And then actually matching that. Again, if you have an ad creative, and you know what the emotions are, and then you can reach a billion impressions a month of those same exact emotions. We know that that’s been proven to drive sales and to drive recall. That’s something that would help align with a lot of industry creatives and the actual creative side of the industry. There’s all this work that goes into creative, and often the handoff from creative to media is really hard, especially if you have more and more creatives.
So, introducing some automation and tools that humans can leverage to help understand the nature of the video content and then make sure that they’re then aligning that message with the right content. That’s where we would like to see this go. Not only because it drives great outcomes, but it is also going to reward the best content. It’s going to reward the content that really evokes emotions. That’s when they’re will be able to justify those high CPMs as they should.
MB: All right. One more question. If you could wave a magic wand and change one thing about the video measurement space, what would it be and why?
FG: I think it would be the big debate currently between panel and actual data. There’s not a lot of people that believe that data, and having scaled back access to information is going to win, but it’s something that you need to be really careful about. I’d like to see everyone who’s offering measurement to actually be able to measure on the same set of criteria. That way it’s on equal grounds and this headache goes away. And it stops being the same thing that everyone talks about year, after year, after year.
That’d be one. And two, we would like to see the video level data incorporated into those solutions, just like we were talking earlier about contextual emotion. If you can introduce those as part of an “always on” part of your buy, you’re not just targeting households and guessing who’s watching the screen. You make sure you’re tapping into mindset, and then you can measure that. We’d love to see video level data incorporated in all those solutions. So that’s a big lift.
We’re starting to work on that, but we are going one piece at a time.
I’ll just start with what everyone wants, which is everyone agreeing on the same dataset so that there is compatibility, and it’s not such a nightmare for buyers. I think my wish is the same as everyone else’s on that one.
MB: All right. Field, you got some exciting news to share with us.
FG: The exciting news that just came out is that Group M is incorporating the IRIS ID into their primary channel of CTV buying. They’re really focusing on leveraging video level data and this new data set as they help clients navigate the investment and shift from TV into CTV. They’ve really honed in on video level data. They recognize the differences that clients are going to have when they go from a medium like television, where everyone’s seeing the same content, to digital and CTV where it’s everyone’s seeing unique content in the home. It’s all on demand. There are billions of potential streams of where that content could be.
We’re very excited about this partnership and enabling our contextual data partners and the brand suitability data partners to help unlock significantly more investment in the CTV ecosystem and all the IRIS enabled partners that we work with.
MB: That’s huge. Congratulations.
FG: Thanks.
MB: Well, Field, I’m grateful for your time. Really enjoyed the conversation. I know our audience is going to love it as well.
FG: Great to be here.
MB: All right. Thank you.
See the rest of the Screen Wars Thought Leader Interviews here!
Field Garthwaite is the Co-Founder and CEO of IRIS.TV, a video data platform that enables better viewing experiences and advertising outcomes. Since inventing the core technology behind IRIS.TV, Field has led the company to engineer the most open video data ecosystem in the world, enabling its partners to build scalable solutions on top of its platform, including video-level contextual and brand-safe ad targeting, planning, verification, and measurement solutions. Today the company works with thousands of media companies, data, streaming, and ad tech partners around the world to power meaningful connections between brands and consumers. Prior to IRIS.TV, Field worked in digital distribution, data architecture, and video for companies including HBO, Universal Pictures, Rubicon Project and Jukebox TV.
Cross Screen Media is a leading CTV activation managed service for marketers and agencies, built on a proprietary technology platform that enables advertisers to plan and measure advertising across Connected TV and audience-driven Linear TV at the local level. We seamlessly fit into existing workflows to help agencies scale, differentiate and deliver high-impact campaigns for their clients. For more information, visit CrossScreenMedia.com.
