How Team Liquid seized a competitive edge in esports with SAP analytics and AI

In recent weeks, we’ve seen plenty of AI backlash, and for good reason. Assessing the hype factor is a worthwhile undertaking, and I’ve done my share, including the occasional snarky asides. 

But one thing I’ve never denied: AI absolutely has the capacity for enterprise scale. When you think about past hype trains that lost steam – blockchain quickly comes to mind – questions of enterprise scale were never resolved.

That’s why scale and performance are nowhere near the top of the AI concerns list – but we still need proof points. At SAP Sapphire, we got some surprising ones, via Team Liquid’s SAP experience. In Team Liquid’s case, SAP’s AI infrastructure can process half a trillion data points in real-time. Yeah, that checks the scale box. 

Team Liquid‘s story is surprising in several ways: for one thing, they are not an SAP ERP customer (though they are a Concur Expense shop). And this isn’t an AI photo opp either. In fact, Team Liquid’s partnership with SAP started in 2018, with the goal of bringing SAP’s data and analytics to bear on Team Liquid’s competitive pursuits. For those who don’t track gaming, Team Liquid is an esports organization that regularly competes in game competitions. Fast forward to 2024, and the Team Liquid/SAP partnership has big time stats to share:

10,000 hours of manual work saved by SAP Business AI; 6 million esports matches stored in SAP HANA Cloud, and 1.6 historical terabytes of historical game data analyzed. 

Tracking patterns and metrics – analytics as a competitive edge

But what does this mean in a competitive framework? As per Team Liquid: 

‘Tracking patterns is a very big part of what we show the team,’ Haitham said. ‘For example, here at Worlds, with the new format, we’re gonna have a very short time to scout opponents. Possibly less than 12 hours from the time the game ends. So there’s essentially no time to just go through the last 5, 10, 15 games to track patterns. But what we have with the SAP HANA Cloud is key metrics and a map tool that gives us information on what this team’s style is, what kinds of things we need to focus on, and we go to the match with that information.’

Both Haitham and Jabbz could point to specific instances where the data made it much easier for them, and their teams, to counter their opponents. For Haitham, the data helped Team Liquid Honda take revenge on Evil Geniuses and qualify for Worlds. [Haitham Algbory is Head of Analytics & Data, League of Legends at Team Liquid; Mathis “Jabbz” Friesel is a German Dota 2 player who is currently a coach for Team Liquid].

Evil Genuises? Doesn’t sound like opponents you’d want to mess around with! But hold up – just because you have millions of data points doesn’t mean that you can apply them properly. However, in this case, there is a result: 

‘Evil Geniuses were a team that would usually play around mid lane,’ [Algbory] said. “In the series we played them, they changed that approach to focus more around top lane. After the first game we started noticing that trend, so we tracked their movement, we tracked their patterns, and it was consistent with the second game as well. So I was able to help the jungler understand what should work to spot the opponent early on and where we can expect them to come top lane, and we were able to counter a lot of their map movement. That gave us a big edge to win that series.”

Half a trillion data points – the data demands of gaming analysis

At SAP Sapphire, we learned more. It started with a day one Sapphire keynote call out from SAP CEO Christian Klein. Later, at an SAP AI customer panel for media/analysts, Team Liquid’s co-CEO Steve Arhancet shared how thousands of gaming actions become insights: 

We’re the largest esports team, the world’s most successful – we’ve won more championships than any other team… When you think about winning, there’s a lot of things that contribute to that, just like in basketball and baseball. But within gaming, what’s really interesting is that they’re sitting at a computer. There’s all of these data points, from moving the mouse, to clicking the mouse, to keyboard strokes, to eye movement on the screen, to the comms that are using my headset – all of this needs to go into a piece of software that is organized and then filtered – and then display that information to the players and coaches and managers, so they can synthesize that, and make decisions on what patterns they want to follow, and where to optimize their play. 

Those data points turned into a gaming practice app, running on SAP technology: 

What we’ve been able to do is build this proprietary software in partnership with SAP, to bring those data points to the surface, so that our players and coaches can basically make better decisions than our competition – and that’s what we’ve done. We have League of Legends, which is one of the most popular esports titles today. We built this draft AI tool that allows our players to make great decisions in practice, and that’s yielded to competitions. We were just playing in the League of Legends Championship this year; we kicked it off at the start of the season, and we ended up winning our championship last month. I think a real part of that has to do with this access to information that our competition doesn’t have.

Just as a quick data point, in League of Legends, just the software that we built, it has over half a trillion data points that are being processed in real-time, and then the player is making decision on the fly using that data. 

During our sit-down interview, Team Liquid Partnerships Manager Thom Valks told me that – believe it or not – Team Liquid used to try to do this in Excel: “It was a lot of manual work,” he said, which sounds like a huge understatement. And, Valks acknowledged, the results were more than a bit random. I asked Valks: how has SAP’s analytics made the difference? He alluded to another crucial lesson: when your data isn’t great, your tendency is to fall back on emotion-based decisions. 

We used to have a team of three or four analysts, but there was still a lot based on emotion, so you would see something in the game, and then people would just kind of go off. So being able to get real data insights in a quicker way, and being able to then show that in the dashboard – players have really started to use these dashboards and have them open. That’s really number one. And then number two is looking at competitors. What are they doing? When they are playing alone, practicing in practice matches, analyzing that.  [Author’s note: Team Liquid only uses these AI and analytics tools real-time in practice matches, as per the current rules of esports competitions].

That’s something that before we didn’t even have the capacity, purely on manpower, to do. So with this tool, I think it’s become so important for us, these dashboards and how our analysts, especially, but also our coach and players – are able to access it and get insights.

Valks says their SAP analytics pursuits began with League of Legends, and then into Dota. AI projects began with League of Legends. As SAP’s push into generative AI continues, the tech behind these apps changes. Here is SAP’s explanation of what is under the hood: 

By using the SAP AI Core infrastructure to train a generative AI model that takes data from past matches and puts them into the context of an upcoming game, Team Liquid gets suggestions of the best draft picks and bans to maximize the chances of winning. The solution is built on SAP Business Technology Platform (SAP BTP) and the data is stored in SAP HANA Cloud to cope with the 1.6 terabytes of game data from past games. The result is an intelligent draft bot trainer, an application that runs on SAP BTP, visualizes the predictions, and provides Team Liquid with the current winning probability after each pick and ban. Relying on SAP BTP and SAP HANA Cloud, SAP Business AI provides Team Liquid with the opportunity to further improve its game by leveraging data to make more informed decisions. [Author’s note: SAP Analytics Cloud is also in this mix, for the dashboarding referred to in this article].

The wrap – is AI a threat or asset to esports? 

Volks and I talked about the future of AI in gaming. Will future competitions someday allow AI “co-pilots” into the real-time mix? Will AI’s ability to excel at particular games (chess, Go, etc.) diminish interest in human gaming competitions altogether? Volks doesn’t think so, and I agree. The attraction to gaming is about culture, not super-human machine performance. 

Though I’ve hammered on AI and creativity, I do think AI can play a purposeful role in creative pursuits, gaming included. I just want to see the tools put in the hands of creators, rather than having market conditions dictated to creators by big tech, which is more interested in pure AI scale than the irreplaceable role of human excellence – and (re)invention.

What we’re seeing from this partnership is promising, with AI tools in the hands of human competitors. As other teams catch on, Team Liquid might lost some of their initial analytics edge. But their continued push with SAP into gen AI should help. 

Volks told me that SAP provides them with developers who have actual gaming experience in apps like League of Legends. This allows the inter-company teams to speak the same language; SAP developers participate heavily on Team Liquid’s external Slack channels. We hear a lot about AI from the keynote stage, and we all know why. I hope I’m not the only one who finds this cross-company collaboration aspect just as compelling. So many AI stories this spring were about productivity pilots. But what about creative app building? What about utilizing big data sources in imaginative new ways? Volks: 

These dashboards were very, very customized. The same goes for the AI solution. Currently it’s only a League of Legends project, but what does that look like? What is it supposed to do? What metrics do we think it would pick, to decide what to draft or to pick or ban or whatever? This is very much a collaborative project, I would say, between us and SAP.

White board discussions are energizing, but it’s even better when they are built on results: 

Since the beginning in 2018 to where we are now in 2024, SAP has been really instrumental in helping us win – to a degree that I don’t even know if we could have envisioned at the start. It’s really something else.

And that, folks, is how you score a spot in my event highlights series.