• Streaming’s Time Problem

    Video streaming (OTT and CTV) has rapidly accelerated in terms of adoption and strategic importance for media companies. It has enabled a dynamic shift in how we consume content and changed the model for content production and distribution. At the same time, it has created a host of new challenges—especially when it comes to advertising.

    This rapid growth and consumer behavior shift has also highlighted the increasing importance of these channels as core revenue generators, witness the acquisitions of Pluto, Tubi, Xumo and the high profile launches of HBO Max, Paramount+, and others. The revenue numbers are large and growing. Hulu is on track to do $2.7 billion in 2021 ad revenue alone, Fox expects Tubi revenue to more than double, and eMarketer projects that CTV ad spending will increase 40% from 2020 to 2021.

    Yet ad failure rates are significant. Depending on whose numbers you rely on, ad failures across all types represent up to 40% of all potential impressions, which equates to very significant opportunity costs for CTV/OTT publishers.This problem becomes particularly acute with AVOD Server Side Ad Insertion (SSAI). With live and linear streaming, prefetching can be utilized to make multiple ad requests far enough ahead of an ad pod to fill that pod efficiently. With AVOD, though, the ad requests, stitching, and often the encoding and segmenting all need to be handled in a compressed period of time between when the user requests a video and playback start. For AVODs, this process is a key contributor to longer time to first frame (TTFF), which often has a significant impact on stream abandonment and therefore revenue generation.

    To understand the challenge, imagine that an end user has just selected to play an AVOD video. Before that video begins to play on their screen, the player will initiate a cascade of events triggered by a call to the media owner’s ad insertion vendor, who will in turn call the media owner’s ad infrastructure for advertising to fill the requisite impression avails in the stream. Once all of the ad decisioning has been completed, the ad insertion vendor will download, encode, segment and stitch the ads into the manifest for the stream.

    It’s only at this point that playback can begin. Since startup delays need to be avoided, it’s critical that timeout limits be imposed on all of the above ad infrastructure parties. Oftentimes these timeout limits are not long enough to get a valid response, or not enough data is passed to the buyers to make a decision with transparency about what they are buying. Beyond this, ads often don’t respond when called by the ad stitcher. As a result, the ad insertion partner will often serve a house ad, or truncate or eliminate the ad pod. So, quite often, the ad load in a given stream is not ideal in terms of revenue generation.

    The ideal solution would be as close to the “edge” as possible, making adjustments based on the actual device environment to enable time shifting of the ad decisions. Using processes on users’ devices to solve this problem can address the limitations of network-based solutions by enabling decisions to be made and effectuated after the manifest has been delivered and playback begins. Aside from scenarios where pre-roll ads are being run, there is a significant time window to remedy any problems that occurred with the initial round of ad calls/decisioning. Taking advantage of this window, when the video is being played, could retrieve better results (fill, price, relevancy, targeting, etc) from the publisher’s ad ecosystem.

    Of course, Client Side Ad Insertion (CSAI) exists. Yet while still in use (about 40% of the OTT market uses it) CSAI has its own problems. These commonly include; latency during player (video to ad) switchover, buffering during ad load, video vs. ad quality, ad blocking and more—all of which impact the user experience and revenue. As such, the trend is heavily steering towards SSAI, which has a lot of operational advantages. As such, I’m advocating for a new approach that leverages the current SSAI process which enables time shifting for ad decisions.

    User expectations and technical processes place time pressure on streaming systems. When a user requests a video, they want it to start as quickly as possible, which is why the current AVOD SSAI approach is taken. However, the time pressure and speed at which ad decisions must be made creates scenarios where a significant amount of revenue is left on the table. It’s not difficult to imagine that this opportunity cost could easily exceed billions annually. Getting more creative about how to maintain rapid startup time and how to approach the ad decisioning process is essential for continued consumer adoption/satisfaction and to ensuring maximized revenue opportunity.