Contextual targeting in OTT (over-the-top video) sits at the juncture of two major digital industry trends: the rapid growth of digital video consumption since the beginning of 2020, and the search for strategies to deploy and scale ad campaigns without relying on third-party cookies.
That statement alone assumes a whole lot of background knowledge -- at least a few years' worth of industry trends. Let's back up and explain how we got here, and what we mean when we talk about contextual in OTT.
What's the Difference Between OTT and CTV?
Over-the-top video (OTT) and connected TV (CTV) are often referred to interchangeably -- which can be confusing, because they're not the same thing. OTT refers to the video content -- video streaming through apps. CTV refers to the hardware -- an internet-connected device for streaming and viewing video on a TV-like screen. When we talk about OTT, we're talking about services like Netflix, Amazon Prime Video, Hulu, or Disney +, to name just a few. When we talk about CTV, we're talking about devices like Smart TVs, Chromecast, Apple TV, Roku, or various game consoles that can stream video (e.g. a Playstation or Xbox). You can stream OTT video via CTV -- but you can also stream OTT via mobile or desktop, which are not CTV devices.
What's Changing in OTT?
The cord-cutting trend is a familiar one -- viewers have gradually been ditching linear TV subscriptions in lieu of OTT services for years. But the stay-at-home mandates during the Covid pandemic accelerated that trend, as viewers had more time to spend watching video content and were looking for ways to save money. Over the course of 2020, the amount of time viewers spent watching streaming video increased by 44%. Viewers are spending even more time streaming video in 2021 than in 2020, while their time spent watching linear TV will be down 7% YOY. Cord-cutters typically don't end up re-tying those cords in the future.
This makes OTT a hot commodity, and as such, 73% of ad buyers who bought CTV inventory this year say they've shifted some of that spend from linear to CTV. Digital video (which includes OTT content) will account for 56% of all video ad spend in 2021.
How Does Contextual Targeting Work in OTT?
Contextual targeting differs from audience targeting, in that audience targeting relies on data indicating a user's interests, behavior, and demographics; while contextual analyzes the content itself to align ad campaigns with relevant environments. With text, this is pretty straightforward: An algorithm scans pages for keywords relevant to the campaign. More advanced contextual solutions can analyze text for sentiment and tone -- not just what the language means, but what meaning it would convey to a human user.
Video requires a different technological approach. The amount of text available to analyze algorithmically is quite limited. There's some text to be found -- metadata and closed captions, for example. But to do contextual in video, the algorithm needs to recognize shapes and patterns. This is extremely important for ensuring brand safety. An advertiser might want to avoid buying around a scene with nudity or violence. A soda brand might want to avoid a scene where characters are discussing quitting their soda habits. A travel brand might want to be aligned with a sunny beach scene -- in which case the algorithm would need to recognize a sunny beach.
However, contextual analysis of images and video is a much bigger technological lift than of text. It requires extensive machine learning to teach the tech to recognize patterns. Even more advanced tech capabilities are required to analyze sentiment -- to determine, for example, whether a scene is comedic or tragic or ironic. That's where computer vision comes in, to analyze data and perform image recognition. Computer vision also introduces efficiencies that help contextual targeting scale in video, and helps advertisers discover brand-safe, relevant video content beyond their tried and true inventory sources.
Getting these contextual signals from video content at scale is a significant development, because OTT services -- which are commonly walled-garden environments -- make their own decisions about what data they want to share with advertisers. One benefit of contextual is that video publishers and streaming services will be able to share more data about their programming, without relying on data from their viewers themselves. Sharing more information on the content where their ads are running will encourage advertisers to increase OTT/CTV spending.
How Is Contextual Good for OTT?
In a few words, audience targeting has always been a challenge in OTT, because the CTV environment doesn't support cookies. OTT usually relies on IP data for audience targeting, and IP data relies on signals from the user. IP targeting won't be a reliable strategy for long: As the industry moves toward giving users more data privacy controls, we need to assume IP data will be off the table around the same time Google deprecates third-party cookies, in 2023. The digital ad industry will need a new playbook, and contextual -- already part of publishers' and advertisers' toolkits -- is a clear solution to this pressing problem.
Also, the scalability of OTT contextual targeting just got a boost from the IAB's newly released Content Taxonomy 3.0, created for CTV, podcasts, and mobile apps, including games. The updated taxonomy breaks down the "news" vertical into more specific categories (i.e., separating news from op-eds), and adds more "entertainment" categories for CTV. This aims to better ensure brand safety in digital video, and to address another longstanding pain point in OTT/CTV -- lack of industry consensus on content taxonomy [https://digiday.com/media/building-that-ecosystem-contextual-advertising-begins-to-sprout-in-ctv-ott-ad-markets/].
There's also a big contextual opportunity in in-video ad formats -- animated overlays below or aside the video, rather than in the video stream itself. The format is basically analogous to the logos and banners viewers recognize from linear TV, but adapted for OTT/CTV. It opens up more inventory and a new revenue stream for publishers, without competing for attention with in-stream video ads. When an advertiser adds an in-video ad to one of their own ads in the video stream, they can essentially create a video take-over. When used on their own, in-video ads greatly reduce the expenses of advertising in OTT/CTV, because they don't come with the high production costs and timeline of traditional video ads.
The contextual targeting partner also plays a role in collecting and sharing video insights. Video URLs are not usually passed through the bid stream, so the contextual partner needs to access the publisher or platform's CMS to pull valuable contextual data for the buy side.