How to Scale Addressable Advertising with Low Latency for Live Streaming

Written by: David Springfall, Founder and CTO of Yospace

Under David’s technical and strategic leadership, Yospace pioneered server-side ad insertion and established itself as the global leader in OTT stream monetisation. The Yospace technology supports full one-to-one personalisation and provides a true broadcast-quality user experience for live and on-demand.

AWS Builder Community Hub

Today, Yospace services major networks and providers throughout the world, including AT&T Entertainment Group in the US; Sky, ITV, BT Sport, Channel 4, STV in the UK; Medialaan, Sky Deutschland, ProSeibenSat, Bonnier Broadcast and Telia in mainland Europe, and Seven, Nine and Network Ten in Australia.

Delivering addressable advertising in live streams is critical if video service providers are to remain competitive as viewing shifts from traditional linear to online.  To deliver addressability at scale for live is challenging, and many of the complexities underlying on-demand server-side ad insertion (SSAI) and delivery are amplified when you bring in low latency streaming.

Consider the demands that a major sports event imposes on the ad ecosystem when all viewers – each receiving a personalized streaming session – go to an ad break simultaneously.  In this scenario a “just-in-time” strategy to ad calls would probably not give the ad server enough time to respond.

The length of time available to resolve all those ad requests is dependent on the video segment length.  If you use just-in-time ad resolution, what we have found is that 90% of the ad requests to the ad server occur in about 1.5-times the segment duration you have configured for your stream.

Shorter segment sizes help reduce latency by taking up less buffer time on the encoder and player-side when delivering content. For instance, Akamai’s platform supports two second segments as opposed to the traditional four- or six-second segment, translating to an approximately 10 second end-to-end latency (1-2 seconds behind broadcast).

However, smaller segment sizes mean there is less time for the system components to respond to or recover from any issues. In the case of SSAI’s just-in-time ad resolution, the primary ad server (ADS) and demand- and supply-side advertising platforms (DSPs and SSPs) must process your entire audience in less than three seconds, as opposed to 8-10 seconds (per the industry norm for latency with four- or six-second segments).

Unfortunately, live events are inherently unpredictable in terms of the audience interest and the advertising break patterns. For example, we often see relatively unremarkable sports matches receive a surge in traffic when a dramatic match goes to extra time which also gives additional break opportunities.

In these unpredictable situations, the ADS can be flooded with ad calls, overwhelming the SSP and, of course, any downstream DSPs.  In such an overloaded situation, far more ad requests are sent to the SSP than expected and may end up timing out. To combat this scenario the use of an effective prefetch system is critical.

Prefetch is the means by which ad calls can be made in advance of a live ad break happening.  It is dependent on the ability of the SSAI provider to be able to “look ahead” and detect the duration of a future ad break.  With such visibility, the SSAI platform can begin making requests to the ADS early and spread them over a longer period of time, sufficient enough for onward calls and responses to and from SSPs and DSPs.

By the time the ad break arrives the ad decisioning process has been completed and all that remains is for the SSAI system to stitch the new ads into each viewer’s stream – a process that is far more lightweight and scalable.  The result is a true TV experience for the viewer and substantial new ad revenue opportunities for the video service provider.

Delivering addressable live streams with low latency at scale is challenging, but with an effective SSAI and prefetch system in place, there is every reason for providers to be optimistic.

Read more about prefetch in our “Go Live” white paper, co-written by Akamai, SpotX and Yospace:


*** This is a Security Bloggers Network syndicated blog from The Akamai Blog authored by Akamai. Read the original post at: