Last month, the United Soccer League (USL), the second tier of club soccer in the US, teamed up with WSC Sports to efficiently create video content using artificial intelligence (AI).
For the USL and its 33 clubs, which perhaps don’t benefit from the same resources as their major league superiors, the ability to leverage machine learning technology to produce automated video highlights is a priceless benefit. The league is now not only capable of putting together easily customisable highlights packages within minutes of the final whistle, but also reels of individual player performances.
The key, of course, is to reach more fans and increase awareness of the league both in the US and around the world - and if getting more engagement was the plan, then the partnership is already bearing fruit.
A recent clip of a rather comical last-minute goalmouth scramble from North Carolina FC’s recent 1-1 draw with Charleston Battery has been posted, retweeted and liked thousands of times over by global media giants on Twitter, while a video of Phoenix Rising striker JJ Johnson scoring a stunning overhead kick against the Tulsa Roughnecks left jaws dropping around the world.
It would appear, then, that AI is already helping the USL engage with more fans than ever before. SportsPro caught up with Vincent Wiskowski, the league’s director of digital projects and emerging technology, and Shaka Arnon, head of US operations at WSC Sports, to find out exactly how an automated highlights platform works, and to discuss the opportunities it creates for clubs that might have previously struggled for exposure.
How did the partnership between the USL and WSC Sports come about?
Vincent Wiskowski: We were trying to figure out how we could organise our game footage and continue to operate in a way that – as we continue to grow so quickly with our number of teams and games - we can provide and create video content in a way that doesn’t become insurmountable.
At the start of the 2017 season, we were already at over 450 games and were going to grow by about 20 per cent for this season, so that presents a number of challenges when you’re trying to create video for social and you have 14 or 15 matches every Saturday. So from a staffing perspective it became important to handle that, and the key thing was that there was a big gap between the metadata and the information we had about our games in terms of what was happening on the field, but then we didn’t have a way of combining that with the actual video component.
I was told that WSC has been powering the National Basketball Association (NBA) in everything that they do on the video side for a number of years, and the NBA is one of the best in sports at utilising its game footage, so it became quite clear that we needed to get a partnership moving very soon.
How exactly does the platform work in terms of simplifying the creation and distribution of highlights?
Shaka Arnon: We combine analysing video, audio and data feeds to identify each and every play – whether that’s every shot on goal, every save or every booking – and create a clip for it by finding a start time that makes the most sense in the context, and then aligning that with the end time. That’s really the secret ingredient of how we use automation to find clips that a human would have created.
That’s a very basic layer of the technology, but let’s say we have 300 clips or plays for a soccer game, we also give the metadata to each clip. This means everything is filtered, creating a search engine for plays, so if someone wants to find every shot on goal by Didier Drogba, they can find it at the push of a button, and they can also filter it by ratings that are measured by the impressions of the crowd and the characteristics of the play.
This is part of the revolution, and the revolution has already started in terms of how rights holders create content, and more importantly how fans will consume that content.
On top of that, we now have the algorithms which allow us to create video compilations from all these clips which can be packaged according to an individual’s preferences as a producer as a fan. This means someone can simply search footage of Drogba’s shots from the last couple of months, and after one click they will have a fully-edited video equipped with transitions, replays, overlay graphics, intros, outros, sponsorships etcetera, and this is now what’s available every day to the USL.
What’s the scale of the operation in terms of monitoring the platform?
SA: We don’t want to send junk content at any cost, so we have a confidence level that is embedded in the system which is all through automation. When we adapt to each sport, the system learns and gets better through machine learning, so the confidence level goes up accordingly. If the confidence level is below a certain threshold, it goes to our assurance team to make sure everything makes sense, but otherwise it’s all through automation.
This has been referred to as a bit of a game-changer for the USL and its clubs. What are the advantages which weren’t available previously?
VW: We have 33 clubs around the US, and a significant way we work with them is by providing a number of services to help them be more efficient and provide better experiences for their fans.
One of the issues that we had was that our clubs didn’t have the resources from a staffing perspective nor from a technology perspective to appropriately leverage their gameday footage. So during the 2017 season, for example, less than five per cent of the clubs were using match footage on any of their digital channels at all. The reason for that was that they were either lacking the video staff to do that, or more importantly, they didn’t have a way of easily pulling and creating video content in a way that they could do it quickly.
So this partnership wasn’t just a league benefit that was going to help us here at the USL headquarters, but it was something that we wanted to present as a tool for all clubs that they could opt in for, and then benefit from using the same system.
So in 2017, one of our clubs might have had zero match footage during the game, whereas now, in this season, they are doing in-game live tweeting of the best moments, so chances are getting posted in real-time. They’re also doing a half-time highlights reel, they’re doing a goal that they can overlay with a sponsor graphic, and then they’re rolling out a full highlights package within ten to 15 minutes of the game being done, which they can then distribute to their local media partners and players, so they’re able to open up their distribution channels much more than ever before because it’s all done through an automated system.
It’s not only a great way of creating video, but it also opens up a new sponsorship inventory that never existed, and more importantly, being able to distribute and grow their brand primarily through their players, because you can have the system create post-game highlights for every player that takes part in a match.
Strategically speaking, how does this move tie in with how the USL wants to be presented?
VW: We have a very big reach, but it was always at that local level, so we’re very relevant locally in the markets that we exist. Unlike some of the top leagues in the world that are known internationally, we’re just different in that sense.
Everything we do has to be about driving local interest, and more importantly the people that care around the world for a particular reason, particularly about the involvement of players and things of that nature.
Instead of creating a one-off reel of the best goals and expecting it to perform very well and reach a lot of people, we’re now making specific targeted pieces based on the audiences that we know.
So this is a logical step for us because it allows us to be very specific about how we use our video. Instead of creating a one-off reel of the best goals and expecting it to perform very well and reach a lot of people, we’re now making specific targeted pieces based on the audiences that we know, so we can build relationships with media in specific markets and provide them the video that’s going to support the storylines that they’re going to want to tell.
We’re trying to be relevant everywhere. Now we really have an engine that powers all our video content, so the way I see it is now whatever digital platform we want to be relevant on and have a presence, we can do that because we have the ability to create at scale.
Do you think we’ll see more and more leagues adopt AI-fuelled highlights platforms such as this one going forward?
SA: There’s no way that automation is not a big part of the future of sport.
This is part of the revolution, and the revolution has already started in terms of how rights holders create content, and more importantly how fans will consume that content. Fans will get used to getting more and getting things that are more customised.
What I always say is that we’ve gone from linear broadcasting to video on demand, where you still have a very limited choice of videos to see, but now I would call it ‘video on command’, where you will see precisely what you want, when you want it and how you want it. So whether you want it in the form of a story or in the form of a pop-up message, it’s going to be aimed at you in one video featuring a recap of all the sport you want to watch.