NFL IT Leader on How the League Uses AI, Data Analytics

The National Football League is a prime example of how an organization has adopted artificial intelligence and data analytics in numerous ways to enhance fan engagement, improve player health, and boost the game’s business.  

A tool called Digital Athlete lets the league predict injuries using AI and machine learning (ML). It uses AWS software to run millions of simulations to predict injuries that might happen in the middle of a game. Meanwhile, a program called Next Gen Stats (NGS) lets the league collect data by visualizing the action on the field to boost fan engagement.  

Big Data Bowl, hosted by Amazon Web Services, is a crowdsourced competition that challenges users to develop data insights that enhance the fan experience and improve player safety. Participants have developed ML models that can identify blockers and pass rushers on every pass play, study the dynamics of pressure, and spot particular blocker-rusher matchups.  

Although all teams are advanced in their use of AI and data analytics, the San Francisco 49ers have stood out for their ability to consolidate disparate data silos and future-proof its services such as Qumulo hybrid-cloud data storage. 

At the AWS Summit in New York City on July 10, InformationWeek sat down with Paul Ballew, the NFL’s chief data and analytics officer, to learn more about how the NFL collects real-data insights and how CIOs should approach AI and data analytics.  

Related:How Tech Turned an Emerging Generation of Sports Fans Into Bettors

(Editor’s note: This interview has been edited for clarity and brevity.) 

How is the NFL transforming fan engagement through real-time insights and data and making the fan experience more personalized? 

It’s at the core of the journey we committed to a couple of years back. It’s our ability to see and know and then engage with fans meaningfully. It’s a systematic approach. It’s not easy to do. We are working with all 32 clubs, so we’re one of the first sports leagues to bring all the fan data together. We spend a lot of time with fans to be very transparent and have permissible use [of] their data. We’re then able to for the first time bring it all together so we can properly associate who you are and what’s important to you. We then deploy that through a variety of ways. We make it available to clubs so they can interact with their fans intelligently. We use it at the league level because the league does quite a bit of marketing and outreach as well, and then we’re using it for our direct-to-consumer business. In terms of technology, it’s across the entire data and analytics ecosystem. We’ve made substantial investments on the data side, because if you can’t bring the data together, if you can’t make sense of the data, if you can’t deal with all the legacy challenges, you’re never going to be able to scale.  

Related:NHL, Presidio, and a Transformation Power Move with Hybrid Cloud

And then our ability to interact with fans is being powered by everything we’re doing on the analytic front, creating audiences, creating offers, learning about the responsiveness to those offers and those interactions. Sometimes it’s information about things like tentpole events, the draft; other times, it’s actual offers. It could be ticketing, could be encouraging you to attend, could be jersey sales, shop sales. So it’s been a fascinating journey. It’s the fact that sports is now jumping into this really shows the power of the world we’re living in, which the direct-to-consumer world came front and center during COVID and is now also relevant for every industry. If you don’t connect with your customer base and, in our case, the fans, they’re going to be disintermediated.  

You previously served as vice president and global chief data and analytics officer at Ford Motor Co. What type of data analytics strategies have you carried over into the NFL from your time at Ford? 

There are a lot of similarities in the sense that when you think about what you do, if you’re a chief data analytics officer in an auto company, you’re supporting quality; you’re supporting product development, manufacturing, purchasing, and safety. We have the support ecosystem for all of those functions. Similarly at the league, our responsibility is to help the league across all of those dimensions as well as player health and safety, marketing, media optimization, officiating analytics, and assessing rule impacts. The core principles are pretty similar.  

Related:NFL’s Minnesota Vikings Draft LogicMonitor for Reliable Video Access

You have to make sure you are able to bring the data together, make sense of the data, and distribute it into a decisioning environment where people can use it. In manufacturing, that was very mature in terms of the plants themselves, where the actual power data analytics was embedded in the day-to-day operations of a plant. We’re not quite there at the NFL because we don’t necessarily have a line running day to day. But similar methods, techniques, data integrity, analytic tools are powering what we’re doing to help optimize the schedule, or to learn insights on player health and safety about impacts of equipment changes or rule changes. So the mindset and the analytic thought process are very similar. The data challenges you face are very similar. The workflows are different in terms of how it eventually gets distributed, because in a manufacturing plant or a retail, there’s this constant churn of activity, whereas in the NFL, there are games that are played every weekend. There’s the schedule that you have to do; there’s optimizing marketing because you’ve got the draft coming up. So it doesn’t have the same degree of frequency, but a lot of similarities in terms of methods and capabilities.  

How will generative AI transform sports analytics, particularly in the NFL? 

GenAI can be a key engine powering what we’re doing in terms of the one-to-one program, Unified View. So lots of different applications. I don’t know if I’d restrict it to GenAI because when you think about artificial intelligence, yes, GenAI is one category. But machine learning is also within the AI family. And machine learning models are the oxygen of what analytics organizations use. It’s our primary methodological approach to any business problem. 

You mentioned about player health and safety, how are you using data analytics and AI to collect insights in that area? 

It’s a big area for us and a large part of it is about good data, data collection, making sense of data. That includes systems like our electronic medical record system. So it’s a very systematic approach. To me, it’s one of the best examples of the role of data analytics, because a big part of that approach is how are we gathering the ingredients, which is the data, and ensuring the data has the integrity, and you can make sense of it, versus just focusing on the advanced technology. You’ve got to be able to have the ingredients to make a good soufflé. And so on the player health and safety side, we spent a lot of time focused on that. It informs so much of what we’re doing. The kickoff [rules] change we’re going to have for the 2024 season is a good example of that. It can involve equipment in many cases, like the Guardian Caps we use in practice, which reduce concussions, head impact, and head trauma; very important work we’re doing on footwear; and work we’re doing on surface and climate to see what controllables you can impact to improve the overall situation for player health and safety, which is two things: It’s frequency and severity. So you want to lessen frequency, and you want to lessen severity. All of those things fit together in a holistic approach to how we can continue to improve the overall safety of the game, while still maintaining the integrity of what everybody loves about the game. It’s action packed. It’s intense. It has a physical dimension to it. 

So you're always trying to find this right balance, and kickoff is a good example. We do a lot of work around impact, what it would mean to the game, what it would mean to scoring, what it means to offense, what it would mean to injuries, and then you put all that together and drive a rule change. Then after you implement it, we’re focused on measuring it to see if the outcomes we were planning to achieve or hoping to achieve we actually do achieve. 

What types of advanced metrics do you capture to help improve in-stadium experiences?

There’s a lot going on in terms of clubs experimenting with how to use data with very limited latency to optimize concessions or traffic flows or other things. We have a couple of clubs that are on the ragged edge. My hometown Lions have a phenomenal operational war room they run during the game themselves. And it is real time for security purposes, for concessions, for traffic flows, for game flows. They’re using it almost instantaneously to change how they’re operating the stadium before the game, during the game, and after the game. It is amazing to see how far they’ve come and their ability to make sense of the data coming in and being able to act upon it very quickly. You see more and more clubs doing that. And one of the things that has been very gratifying for us because we’re enabling the clubs to be effective and giving them tools and so on is that clubs are doing this on their own. They’re taking what we’re doing, and they’re going down paths that we love. I can’t say we directly said go do this. They just went in and built upon it themselves. The Lions are a good example of that. The San Francisco 49ers are doing some incredible things in their Game Day experience. There are other clubs as well and it’s very impressive. 

What is the Big Data Bowl, and how does it generate insights for the game? 

We have an activity every year, which is more of an open-source competition. Think of it as Kaggle competition where we have a particular problem, and we let people come together and solve the problem. And part of it is we make the NGS data available because we want people to find unique insights. The world we now live in allows us to tap into the intellectual horsepower of a whole bunch of people. The big data bowl is part of that, as well as other things we do with universities. We have university partnerships now in my organization, which are equally powerful.  

What advice do you have for CIOs and other C-suite leaders on how to collect real-time data using AI? 

Understand that this is an enabling set of capabilities that needs to be interwoven into what you do, how you do it, where you do it. And my advice to everybody is think about it in that regard. Don’t think of it as a foreign species. Don’t think of it as something that is so unique that it has to be put off on its own planet. Just about every company that did that 15 years later said, how do I integrate it back into the mothership, and you went through all the organizational consternation, pain, and suffering associated with it. We really need to think about it that way.  

And then always anchor back on your mission as a business and what you’re trying to do. Yes, this will create new business opportunities. But at the end of the day, you are who you are. And the capabilities that are associated with this come in two buckets: They help you drive the efficiency of your activities; and secondly, they afford you the opportunity to connect with your customers better. If you embrace those two things, that’s what this enables you to do, your odds of success are going to be higher.  

There’s a change management component to this as well. You can generate insights, you can do great analytics, but the business partners ultimately have to leverage what you do to make better decisions to change something in their process or the workflow. And so we spend a lot of time in this day and age with our business partners to help them and to make sure we understand what issues they are wrestling with, and we have context. And then in turn, we can help them leverage what we do.