Contextual Revolution: Rohan Castelino on how the IRIS_ID Unlocks New Potential in CTV Ad Targeting
This is the second of five Innovator Spotlight articles from our new Special Report, The Contextual Revolution, Five Companies Rewriting The Rules Of CTV Advertising. You can download the report for free thanks to IRIS.TV and our other sponsors.
“We want to make it easier for more advertisers to enter the Connected TV market and help them drive superior return-on-ad spend,” says Rohan Castelino, Chief Marketing Officer at IRIS.TV. “The IRIS_ID enables this by providing a consistent, privacy-safe identifier for video content, allowing advertisers to target and measure campaigns across all platforms seamlessly.”
ALAN WOLK (AW): What problem are you trying to solve with the IRIS_ID?
ROHAN CASTELINO (RC): One of the biggest impediments to CTV growth is fragmentation. Consumers watch streaming video on hundreds of apps and FAST channels across their preferred smart TVs and connected devices. The same content can be found on different services but is often categorized according to the metadata standards of each provider. The data in the bidstream is, therefore, incomplete, inconsistent, and, more often than not, incorrect. This makes it difficult to plan, target, and measure campaigns across supply sources before considering factors like signal loss and privacy regulations.
This increases the barrier to entry for more advertisers including small and medium sized businesses (SMBs) that make up a significant proportion of global ad spend. We want to increase the total addressable market for streaming advertising and we believe the best path to achieve this is through video-level contextual targeting enabled by our content identifier, the IRIS_ID. When a video has an IRIS_ID, it can be targeted and measured wherever it is watched, regardless of the app, channel, or device it appears on.
AW: What is the one misperception about IRIS.TV and/or the IRIS_ID you’d like to clear up once and for all?
RC: The biggest misperception of IRIS.TV is whether we are a data provider or a data marketplace. We are the latter, but it’s easy to understand the confusion as the marketplace is enabled by the IRIS_ID, our universal content identifier. The misperception about the IRIS_ID is that it can be confused with a person-based identifier. The IRIS_ID is privacy-by-design, which means it is only associated with content information and contains no PII or PII-derived data.
For over a decade, we’ve worked with global publishers to build AI solutions for content discovery and monetization. We started the business focusing on recommendation systems, the first wave of AI as we know it today. While onboarding our publisher clients, we ingested every video’s source metadata and normalized it into a standardized taxonomy. We assign this metadata to the video’s IRIS_ID.
When contextual data companies sought to extend their solutions from display and YouTube to open web video and CTV, they ran into a major roadblock: access to the actual video content. In display advertising and on YouTube, contextual data companies have relied on page URLs as the point of access for their solutions since page URLs are publicly accessible and persistent content identifiers. The challenge is there is no video-level equivalent to the page URL in streaming.
IRIS.TV solved this because we integrated directly with publishers and assigned IRIS_IDs to all their videos. We extended the utility of the IRIS_ID from a content registry to a common signal for data companies to access content and create contextual, emotional, and brand-suitability data at the video level. We effectively built the largest content graph and contextual data marketplace for streaming media.
Because we popularized video-level contextual targeting and made it into a scalable solution for CTV, we are sometimes confused with the data providers that actually analyze the content and create the data. Our data marketplace includes companies like Captify, GumGum, Illuma, Kerv AI, Mobian, Pixability, Reticle, Samba TV, Sightly, Silverbullet (4D), Silverpush, Spot Runner and more.
AW: Why does the IRIS_ID include data from so many partners that offer different types of contextual data like emotion and brand suitability?
RC: The speed of innovation in AI is remarkable, and it's important that the market has access to all types of data. Contextual data like the IAB taxonomy is the foundation for establishing relevance, but there are so many more insights to be gleaned about the content we watch. Brand suitability standards vary among advertisers so it’s important that each video is accurately categorized. Otherwise, advertisers are forced to take draconian measures like blocking all news publishers instead of just specific videos. Like brand suitability, emotional data is an essential ingredient to finding relevance. An estate planning service may prefer to appear in content categorized as “respectful” rather than “trendy.” A fashion brand may prefer to reverse the scenario. It’s our job to make sure that all these solutions can go to market as fast as possible so that buyers and sellers can maximize the value of every streaming video.
AW: It has been said that video-level contextual targeting is real-time targeting. What makes real-time targeting more powerful than other traditional, data-driven methods based on historical behavior?
RC: Real-time video-level contextual targeting in CTV is more powerful than targeting with historical behavioral data because of the unique attributes of the medium.
While consumers stream video on all devices, they typically use their laptops to search and their phones to shop and scroll. That’s why advertising such as search, retargeting, and sponsored posts work well on phones and laptops, because they complement the active, engaged consumer journey. When we watch CTV, even if we begin the journey on-demand, we are passively engaged. We are likely co-viewing with a companion while scrolling on our phones. With all these distractions, the one signal that has the greatest fidelity is our mindset. Eye-tracking research has shown that CTV viewers are four times more likely to watch, remember, and have a favorable opinion of a brand when ads are relevant to the topics, themes, and emotions of the specific program they are watching.
When it comes to data, a laptop or phone is assigned to a person, and a smart TV is assigned to a household. Households often have multiple people with different preferences streaming content on apps that don’t require logins.
Video-level contextual targeting is a great way to achieve real-time relevance because it matches ads to the viewer’s mindset. In addition, it increases the probability that the person watching is the appropriate target for the ad.
AW: Do you see contextual targeting becoming the default methodology for TV advertising?
RC: Contextual targeting is about getting as many viewers as possible to pay attention and watch commercials. This has been the foundation of TV advertising since its inception. In digital advertising, targeting based on identity and behavioral data has offered advertisers greater precision and accountability but in CTV the user experience, fragmentation, and signal loss have fundamentally changed the paradigm. The only way to rebalance the scales is by allowing advertisers to know what viewers are watching in real time and match ads to content based on contextual relevance, emotional resonance, and brand suitability. The key to this becoming a default methodology is the adoption of the IRIS_ID. We’re making progress as over 60 million videos currently have IRIS_IDs, resulting in over 600 billion monthly ad requests, and we are working to increase penetration across the open web and CTV.
