Introducing BrandDiscovery: Shaking up CTV with GenAI-powered contextual targeting

Streaming has transformed the way we consume media. As more viewers embrace ad-supported streaming services and programming, connected TV ad spending is expected to grow 25% to more than $30 billion this year.

Despite increasing ad budgets, the CTV marketplace remains notably opaque. The ability of marketers to target and understand exactly where their ads run has largely remained an out-of-reach goal. This lack of transparency is the root cause of one key pain point for consumers and advertisers alike: in a world with unprecedented levels of streaming content, how do publishers and advertisers best take advantage of programming to ensure ads are relevant, appealing, and ultimately effective?

Think of the program as “setting the stage” for the ad. Traditional, linear television advertising has been the industry’s most premium advertising medium for 80 years precisely because it has enabled advertisers to use programming to forge emotional connections with the audience. It’s time for CTV advertisers to be able to do the same.

Over the past year, our team worked to build BrandDiscovery, a GenAI-powered contextual targeting solution for CTV that helps marketers reap the rewards of finding the right programming for their ads. BrandDiscovery leverages MLLMs’ stunning ability to understand, classify, and apply on-screen imagery, sounds, and dialog to create targetable segments in real time for content genres, brand safety, and even emotional sentiment. Importantly, BrandDiscovery gives CTV advertisers precise control over where their ads appear without relying on the publisher or program data that is generally withheld from the programmatic ecosystem.

The Role of Emotions

BrandDiscovery started with a question. Wurl’s CEO, Ron Gutman, who has long studied the role of emotions on behavior, posed a question to the team: Did ads that were emotionally aligned with content perform better? Afterall, as he explained in his recent blog, emotions are regulated by chemical changes in our brain and not easily controlled. If the emotions of ads and content are not aligned, the ad would have to work all the harder to transition the audience into the best frame of mind for receiving the brand’s message.

Our engineering team set out to answer Ron’s question by evaluating the results of campaigns delivered through two CTV performance-marketing products: Wurl’s ContentDiscovery and AppLovin’s AppDiscovery. We found that the strongest correlation to ad engagement occurred when the emotions of the scene just before the ad break matched the emotions of the ad creative. In fact, data showed that this emotional resonance boosted performance by 2-3x.

Why? Let’s consider an example we’ve likely all experienced: We’re watching a sad movie… still dabbing our tears when all of a sudden the most cheerful commercial we’ve ever seen pops up on the TV. It grabbed our attention, but not for the right reasons: loud music, laughter, over-the-top happiness when we are feeling sad and distrustful. It may be a legitimately great ad, but it wasn’t the right time for us to see it. We’re not inspired by the brand, we are annoyed and skeptical. In this context, instead of earning positive attention, the ad has actually earned negative attention.

As the above example shows, not all attention is equal. While attention can broadly be seen as a predictor and driver of campaign outcomes, there is a real difference in attention that brands earn because viewers feel an affinity for the brand, and the attention that is earned through disruptive experiences. As a previous study by OMD highlights, “Attention drives sales and mental availability. Make sure this attention works for you (and not your competitor).”

Unlocking the Future of CTV Targeting with GenAI

So we learned that emotions played a role in ad performance, but we also learned that timing played a role. Remember, the greatest correlation was linked to the emotions of the scene before the ad break.

Wurl developed its own GenAI models trained against millions of assets to analyze content streams in real time against large scale inventory. Our tech mimics a viewer watching content and identifies the emotional state they would experience. The emotion taxonomy we’ve adopted is based on Robert Plutchik’s wheel of emotions and includes eight primary emotions at the three different intensities (high, medium, and low), which can be combined to replicate the full spectrum of human emotional states.

To make things easy for advertisers, we provide free pre-campaign creative analysis to identify emotions present within their ad creative. This empowers advertisers to match the right ad creative, with the right content, at the right time — delivering positive attention for their campaign. I’m proud to say that with real-time analysis, Wurl’s BrandDiscovery segments are passed in the bid request with zero latency.

Of course, emotion targeting isn’t the only capability of BrandDiscovery. We are equally excited to offer scene-level targeting for genres and brand safety – an approach that stands in contrast to standard program or show-level metadata often used to classify content genres and brand safety. The timely nature of our GenAI-powered targeting strategies ensures advertisers can improve their contextual relevance, as well as minimize brand risk. For example, many programs could be classified within multiple genres. The Office might be a comedy, a romance, or maybe both depending on the scene. Likewise, many programs include scenes that are both brand-safe as well as not brand-safe. A good example of this might be the family favorite, Shrek, a mostly light-hearted animated fairytale which might be traditionally deemed to be brand-safe, but when you look at it scene-by-scene, some brands may prefer to avoid its fight scenes as well as the scene when Lord Farquaad decides to violently interrogate the gingerbread man.

Our Goal Is a More Effective Ad Ecosystem

At Wurl, we are firmly focused on accelerating the shift to streaming. We believe BrandDiscovery is a big step in that direction — one that benefits both publishers and advertisers.

BrandDiscovery continues to protect publisher data, so streamers and publishers can maximize their own direct sales, but it also provides advertisers with access to scalable contextual targeting, giving CTV ads the best possible opportunity to drive positive attention and results. More effective ads will lead to better outcomes for advertisers, and ultimately more CTV revenue for publishers – a win-win for all.

Speaking of results, be sure to check out our Media.Monks case study. When added to their campaign, our BrandDiscovery segments significantly exceeded industry benchmarks for both upper and lower funnel metrics, outperforming the KPIs of their other CTV campaign strategies. Read the full case study here.

To start, we’re making BrandDiscovery available at no additional cost across over 60 billion CTV impressions per month in the US. We’ll be expanding to EMEA soon. To find out more you can visit the BrandDiscovery page or get in touch with one of our team members today.