Chatbots in sports were once just glorified FAQs.
For a while, that’s where they remained. A quick way to figure out the clear bag policy for the game. As the underlying tech improved, the bots became more skilled; they stored more knowledge, understood a bit more nuance, and became more interactive.
But that was all the 1.0 era. Call it the first quarter. We’re well into 2.0 now. They can still do everything the old bots did, answering any question a fan may have on game day, from where the vegetarian food options are located to what time the gates open. The new era is more strategic, more aligned with business objectives, and more woven into every service and revenue-generating department in the organization.
This is the world that Bill Bailey navigates now. As large language models and the ChatGPTs of the world introduced hundreds of millions to unimaginably sophisticated chatbots, the remit for these bots and highly trained ‘agents’ extended from gameday ops to, well, everything. As Chief Revenue Officer at Satisfi Labs, which runs chat agents for about 70% of major league teams, Bailey sees the evolution playing out firsthand in the conversations he’s having with clients and prospects, and the use cases Satisfi is bringing to life.
“Our meetings with our clients shouldn’t be about content. We have that,” said Bailey, who has been with Satisfi Labs since 2019, following stints at sports-focused data management company SSB (acquired by KORE, now part of Two Circles) and the New Orleans Hornets. “It should be about campaigns you’re running. Our agents run campaigns. And so all we need to know is, where are you in your sales cycle? Are you launching renewals in the next month? Okay, let’s start getting the renewal campaigns going. Let’s start making sure I have the latest nudges and I’m pushing people to renew in the conversation, whereas a month ago I wasn’t because it was too early.
“So our conversations have shifted now to be much more strategic and fun about your business. And we’re just an extension of your marketing arm. Not about, ‘hey, we’re not answering where a hot dog is; what should I say?’ Those days are gone.”
One of the biggest accelerants behind what Satisfi’s agents can do, and the generative AI space in general, is the infusion of data. The platforms are connected to data sources, pulling in what a team knows about a given fan (with their authentication and permission) and delivering the level of personalization the modern consumer now expects everywhere from Amazon to Netflix to ChatGPT. And now from their favorite team’s bot.
The fan experience with AI is getting closer to that with a human. A service rep or salesperson knows what tickets you’ve purchased, what level of experience you prefer, whether you’re attending games as a parent with kids or a sales executive with clients, and even what food allergies you have. Bringing a profile and context to the conversation is what separates this era from the last. The capability is ready. But teams are still figuring out which use cases to deploy and hand off to their computerized colleagues.
“If you tell me some more things, like they’re a suite holder or they have a season ticket, then I can do even more,” said Bailey. “All the way to, if you have a data warehouse and the desire is there, we can take that identifier you gave me and we’ll go get what fan profile they belong in, what segment. Have they renewed yet? Things like that.
“So there’s lots there. That’s the next frontier. We’re ready for it. It’s more of the business decisions and user stories being developed by our clients as we speak.”
While uber-personalization is enticing, and often welcomed, there is a fine line between knowing about you and knowing so much about you that it feels creepy. Contextual awareness, like knowing which entrance you scanned in from and pointing you to the nearest restroom, is cool and helpful for many, but concerning for others. It seems good that, because the AI knows you have small kids, it will recommend you stop by the children’s playground outside gate 120, but maybe not so good when it knows their names and hopes Jacob has fun on the merry-go-round.
The capabilities to connect all these data sources are largely there, as Bailey noted, but there’s a human element to introducing this to fans.
“[It] stops short of saying, ‘Hi, Bob, how’s your wife, Jennifer?’ That kind of personalization is available,” Bailey explained. “A lot of people aren’t really ready for that. I’m not sure I even am. That’s the creepier part of it.”
Personalization, fan context, and official affiliation with the team may be necessary parts of the moat for team bots. Because the competition to be the first stop for any query, sports team-related or not, is fierce. More of consumers’ everyday queries are migrating from Google, which at least pointed them to the team’s official site, to whichever LLM they favor. So if a fan is planning their gameday experience, isn’t their first instinct to ask ChatGPT or Claude or Gemini or Copilot or whomever?
That’s not hypothetical. Bailey said teams and leagues are already concerned about web traffic being down significantly, with those queries going to ChatGPT and other models. Still, there are real advantages to having those conversations with the team or venue’s official bot. One is the personalization already discussed, bringing your fan profile and context to the table. But the most important is trust. Skepticism follows any LLM conversation that depends on getting the facts right. The generic LLM may give you a confident answer about the parking situation and the outside food and drink policy, but can you really trust it?
The story is still playing out. Getting burned by a confidently wrong answer may well send some fans back to the source, but old (and new) habits die hard. Plenty of people will tolerate something short of perfect accuracy for the convenience. Maybe the future is reliance on a multitude of specialized, trustworthy models. Maybe it isn’t. But it’s on the official bots, and the platforms powering them, to convince users through a decisively better experience to ditch their daily LLM app. No easy task.
“It is my belief that as those people find out you can get the wrong answer from those models, they go back to the trusted source,” said Bailey. “And our clients are reminding their fans as best they can: we’re the trusted source, come back to us. And by the way, with the help of Satisfi, our experience is very much like ChatGPT when you talk to us. But you’re talking to us.
“So it’s still that really nice, robust conversation, that warm conversation, that intuitive experience — but you’re having it with me, not with some model that scraped the interwebs and God knows what Reddit articles it was referring to.”
Accuracy and trust in information are compelling reasons to use the official team bot. But information is a commodity, and the large, well-capitalized LLMs can close any margin there. So it’s on the teams and platforms like Satisfi to stay a step ahead, doing the things the generic LLM platforms won’t or can’t do as well.
Commerce and service actions are part of it. Paired with the profile connections discussed earlier, an official agent can take a fan from question to ticket, concession, or merch purchase in a few beats, where a general model could only get there with a lot of friction. Bailey brought up something else, though: a novel feature Satisfi introduced with the NHL, which ChatGPT isn’t going to prioritize or invest in.
“…Let’s connect a live feed of every NHL game. And now there’s a game companion module that’s your companion while you’re on the couch watching it, or in the building — but anywhere you are watching the game,” said Bailey, who also noted the bots are trained on the rules of hockey. “And you can say, ‘Whoa, what just happened?’ Because often the play is fast. The referees don’t always stop like an NFL does and announce what happened. It can be quick, you missed it. But what was that? Oh, here’s what happened. That was icing by so and so. That was a goal by so and so. And then, ‘Well, what’s icing?’ And now they’re talking about what’s icing and offsides. And you can actually have a whole conversation about that with our agent today, with our NHL games.”
This isn’t just a traffic and O&O (owned and operated) play for teams, though. Because when you have access to a frictionless, always-on agent that never gets impatient and will assuredly not judge you for a dumb or odd question, you’ll ask anything and everything. And that qualitative and quantitative data is often worth more than any survey can hope to deliver.
It’s not what fans say they do. It’s the actual path they take, including what stopped them and what pulled them through. It can bring up ideas or questions the team never thought of.
Said Bailey: “I often say to our prospects: don’t think of any questions going unanswered. Think about any questions going unasked. And the Predators here in Nashville (NHL), where I live, are a client, so I’ll use that as an example. If they weren’t a client — how important must a question be before I’m going to pick up the phone and call the Predators? That must be really important, because otherwise I’m just going to guess.”
Bailey gave an example from a college athletics client and its softball program. Softball is a non-revenue sport there, admission is free, and fans kept asking the same thing: how much are tickets? That was enough to start the school rethinking whether they should be free at all. Questions are currency. And an official bot collects them at a volume no other channel comes close to.
The passive benefit is ingesting all those conversations. The next step is pre-empting them. Bailey talked about journeys built on a bank of past conversations and outcomes, where the AI learns when and how to ‘nudge’ a fan toward something. Because your ChatGPT is always eager to keep the conversation going, and Satisfi’s bots take a similar ‘and then’ approach. The underlying goals can vary, from getting ahead of your next question to putting on a more purely promotional cap.
That’s the spectrum of nudges. Some come out of what you’re already doing. You’re asking about tickets for you and some friends, and the bot points out a group threshold with experiences and savings attached. You never would have asked and otherwise never would have known. Other nudges come out of what the team’s marketing is prioritizing. Bailey’s example was Korean nachos, a new menu item nobody knows about, worked into a conversation that started somewhere else entirely. Also a nudge, just one coming more from a sales lens than a service one.
That’s the revenue side of being proactive. The service side is quieter, and probably more interesting.
“I think in more and more proactive pushing of that conversation,” said Bailey. “That’s the next level. I know you have a ticket to the game tomorrow, so I’m going to go ahead and give you some information. I’m going to try to start a conversation. Today, that’s in the form of the know-before-you-go emails that people send out, and we’re buried in those emails that say, here’s some information, if you have more questions, ask, click here.”
Bailey continued, discussing the next generation of Satisfi agents and where the space is headed: 1:1, contextual, personalized conversations and services that augment what a resource-limited staff can cover.
“Our newest agent we’re releasing now is a season ticket membership agent, where the endpoint to that agent is specific to that season ticket holder,” he said. “So I can start to know where they sit. I can start to know who their rep is. I can text it and ask questions. Someday, add a ticket to my account because my son’s coming tonight — let’s get a ticket close to me. Things like that. So that’s where we want to head with these, because that’s where the conversations are going.”
Chatbots in sports now reach far deeper into the business than the concessions and venue policies of the earlier era. The conversations are rich with fan profiles, personalized offerings and responses, and whatever campaigns and promotions the team is running that week. Teams and their platform partners need to keep minding the moat, finding the jobs where they can beat the generic models outright, while balancing the service and sales pull on every conversation.
But the chatbot should now be treated as a channel. Heck, it may belong near the top of the list, above touchpoints teams have been resourcing for years, because of how much the agents now know and how much they can do. More sales, better fan experiences, and a running record of what fans actually want. All because there are more conversations happening, and an operation built to analyze and act on them.
The battle of the present day is over questions. Google owned them for years. Then ChatGPT took a big bite. Now, when a sports fan wonders what time the gates open or what just happened on the ice, they have options. Where will they go?
WATCH OR LISTEN TO THE FULL INTERVIEW WITH BILL BAILEY
SKIPPER — TALK TO THE DIGITAL AND SOCIAL MEDIA AND SPORTS PODCAST
