This article by Justin Sampson, our Chief Executive, was first published in Mediatel Newsline on 25th July 2018.
Mo Salah hit the turf, hard, at 8.12pm on Saturday 26th May. It was painful for him, and what followed was painful for Liverpool fans everywhere, including me. So, what’s the point of reliving this, other than the therapeutic benefits of talking publicly about heartbreak?
Well, we know that 4.3 million people were watching the UEFA Champions League final at the moment that Mo Salah was brought to ground. The average audience for BT Sport 2’s coverage of the whole match was 3.1 million, which is a strong result; the channel drew an average audience of just under half this level for the first leg of Liverpool’s victory over Manchester City earlier in the tournament.
A notable feature of the coverage of the final was that BT made the programme available free of charge through its BT Sport app, and also the BT Sport channel on YouTube. The latter was readily available to watch on TV sets, which begs a question: did BARB count this viewing through YouTube?
We did, thanks to a technique called audio matching. This identifies when a TV set is showing content from a BARB-reported channel, such as BT Sport 2, regardless of how it’s distributed. That said, audio matching can’t determine how the content reached the screen. Instead, we have to rely on other techniques to give us clues on how much of the audience might have watched through YouTube.
In the first instance we know that 37% of the 3.1m average audience watched the Champions League Final on something of a time delay. We call this VOSDAL, which stands for viewed on same day as live.
This could be people recording the match and watching it later in the evening. Or, it could be because any delayed viewing of more than 20 seconds is counted as VOSDAL, and we all know that buffering delays of at least this length can happen. I’m not the only one who ignores goal alerts on my phone while live streaming a match to the TV set.
We also know which device delivered the programme to the TV set. In the case of the Champions League Final, just over 62% of the 3.1m average audience were watching through either a Sky, Virgin or YouView set-top box. We expect virtually all of these people were watching BT Sport 2 through their subscription package.
The other 38% of the audience either watched directly from their connected TV, or through other devices that allow casting or streaming to a TV set. We can’t be certain, but a significant majority of these viewers are likely to have been watching through YouTube.
The reason we can’t be certain is that our ability to deliver comprehensive insight is improved with the cooperation of media channels and the platforms they are distributed through. Forgive the pun, but television audience measurement is becoming something of a team sport.
Increasingly, BARB needs the involvement of the media whose audiences we report. For rights owners that distribute programmes through VOD services, this means providing us with non-linear programme assets before they are available to viewers. For BVOD services and other online platforms, this means embedding software code in their apps. In other use cases, such as distinguishing the source of viewing, broadcasters need to embed an inaudible watermark in their channel output.
By way of an example, and in happier times for Mo Salah, there were 125k average programme streams on tablets, PCs and smartphones when Liverpool broke Manchester City’s unbeaten record in the Premier League back in January. We know this because Sky has put software code, to BARB’s specification, in its Sky Go service. This delivers BARB a duration-based metric that’s comparable to our established average audience ratings.
Many other broadcasters have embedded this software code in their BVOD services. This means we can track viewing on devices other than the TV set for the two big television events of the summer: the World Cup and Love Island.
During the group phase in Russia, the most watched World Cup match on tablets, PCs and smartphones was England’s last-gasp victory against Tunisia; the game drew an average of 270k live streams on BBC iPlayer. As you would expect, there was an even higher level of live streaming on ITV Hub as England’s unexpected bid for glory came to an end against Croatia, with an average of 339k live streams.
Impressive as these numbers are, they’re eclipsed by Love Island which has regularly pulled in these levels of live streaming on screens other than the TV set. This year’s edition of Love Island has also captured very high on-demand audiences; the most-watched episode on tablets, PCs and smartphones attracted over 1 million on-demand average programme streams. This sets a new high watermark by some distance, as few other programmes have topped half a million since we started reporting online TV viewing data in 2015.
These examples show how BARB has developed our audience reporting to keep pace with the changing ways that people watch their favourite programmes. There’s always room for improvement, and we have plans to provide more distinction on the source of viewing. To the earlier point, our ability to deliver such insight involves the cooperation of the media we are measuring.
And we also need to provide cumulative multiple-screen audience figures. This is the focus of the next stages of Project Dovetail, which you will hear more about after the summer break.
In the meanwhile, my therapist will be pleased I’ve got through this article without talking about Sergio Ramos. So too will the editor of MediaTel Newsline, because that would have been unprintable.