Handling the opposing forces of programmatic and low latency

Posted on

By David Springall, Founder & CTO of Yospace.


In this post I’d like to talk about the challenges faced by broadcasters in efficiently monetising live sports at scale using server-side ad insertion.

We have broadcasters in five continents monetising live sports using our platform.  During the World Cup last year, we saw a peak audience of 1.6 million concurrent users and only a few weeks ago, in January this year, we saw that record double to 3.2 million – a significant jump in the space of just six months.

So the ability to scale SSAI, and to plan for future scale, is really important but this topic is made complex because there are two opposing forces at play.

On one hand you have the desire for the broadcaster to improve the viewer experience by reducing latency.   

On the other hand, the way in which online advertising is being bought and sold is trending towards automated, or ‘programmatic’, fulfilment.  This makes the user-level ad decisioning much more complex.

Underlying all of this is the fact that the industry as a whole is finding online audiences, especially where major sports events are concerned, are growing at an ever faster rate, and it’s important that solutions deployed can continue to keep ahead of this growth.  

So why is it that these forces are opposing?

The reason for this is that in live streaming everyone watches the ad break at more or less the same time.  In order for us to know which ads to stitch in to each individual stream, we need to make a call to the ad decisioning ecosystem on behalf of each individual streaming session.

Naturally, these calls all take place in a very short space of time.  The lower latency the stream, the shorter the time frame under which this bevy of calls will be made.  As this time window is made even smaller, you need each individual ad decision request to also take less time.  

Ok, so let’s hold that thought.

On the other hand, because the ad decisioning ecosystem is getting more complex, it actually needs more time to resolve ad decisions.  Real-time bidding is required in the process of getting decisions for ad placement, so you’re not making calls to a single ad server, you’re making calls to multiple partners in the ecosystem.  Furthermore, you’re depending on all of those partners to be able to scale to handle these bursts of requests.

Therefore, to deliver the low latency experience at scale for live sports, and to plan for future scale, with a complex ad decisioning ecosystem to maximise the revenue opportunity, it’s necessary to the make ad decisioning calls in an orderly fashion way ahead of the actual break taking place.  By doing this you are no longer hostage to stream latency and how long you can spend, or how many calls you can make to get a decision for each user.

So this is what we do with the Yospace SSAI platform.  We call it prefetch and it allows us to scale to support the world’s biggest live events.  We co-authored a white paper on the topic with Akamai and SpotX.  It’s called “Go Live” and you can download it here.

Schedule a time to meet at NAB to find out more!