Smith Professors Make Sense of March Madness Marketing Trends and AI-Involvement in Bracket Building

Responses from the conversation have been condensed and edited for clarity.

Q: With fans using AI to fill out brackets, is a large language model actually analyzing the games, or is it just parroting the consensus of sports journalists it found in its training data?

McCarthy: “Yes. By that I mean that it depends on what you’re doing. If you ask, “Who’s going to win this next game?”, it’s probably going to give you what sports journalists said in the training data. However, that’s not a very diagnostic question. If you were to ask it to analyze historical upset rates and other questions that allow you to dig in much deeper into what is actually more likely to happen, it can absolutely perform very high-quality analysis. I often think of AI like a mirror. You’re looking into your question, not necessarily into yourself.”

Q: Women’s basketball stars often have higher social engagement rates than their male counterparts. From a marketing ROI perspective, how might the women’s Final Four become a more attractive buy for brands than the men's?

Boyd: “We’re living in the attention economy, which is all about engagement. We have to give credit where credit is due, and the women have done really well in this area. If we look at platforms like TikTok and Instagram, you’re seeing engagement rates two to three times higher in favor of women than men. If I’m a brand, I’m thinking I’d rather have a set of engaged viewers, even if it’s a smaller number, than a larger group with tremendous reach, since a lot of folks let it wash over them. Women’s basketball is here and rising, and it’s a good thing in the sense of opportunities for brands. The guys have to elevate their game on those platforms as we go forward.”

Q: If you ask AI, "Who will win?", it usually picks the favorite. How should a user form their query, or prompting strategy (e.g., "Act as a high-risk contrarian gambler") to help produce the most accurate upset predictions?

McCarthy: “Ask it for the question. There are the 'known knowns,’ the ‘known unknowns’ and the real dangerous spot is the ‘unknown unknowns.’ Those are things you don’t know and that you don’t know that you don’t know them. It’s easy to get hoodwinked in that quadrant of the two-by-two. The analog would be to identify things that drive an upset. Dive into those factors and then see how they map to the upcoming games. I think that step number one of saying ‘tell me what I should know and why,’ will allow you to start asking the right questions in the first place.”

Q: Wendy’s is the “Official Dunks Partner of NCAA March Madness,” with a campaign centered around every dunk on the court being a cue to dunk your fries at Wendy’s. How might this approach, focused on on-court action, benefit a brand compared to solely hedging its bet on a specific player?

Boyd: “Everyone remembers dunks. When it happens, it’s electric. Wendy’s sat down and said, ‘Let’s not anchor on a player, let’s anchor on a behavior.’ The dunking behavior is the phenomenon. What can you dunk in real life other than a basketball? You can dunk your fries in your Frosty. From that standpoint, I think it’s very smart. We look at human nature, there’s a randomness to the tournament where you can get eliminated early, or a star has a bad performance. Suddenly, a brand is anchored on that person, which maybe it shouldn’t have been in retrospect. I think Wendy’s will get great mileage out of it.”

Q: A bracket is binary, but AI thinks in probabilities. Does forcing an AI to make a hard choice in a bracket actually degrade the quality of its underlying statistical reasoning?

McCarthy: “It’s degrading what it’s giving to you from that reasoning. If it believes there’s a 55% probability that one team will win over another, then if you were to ask, ‘Who’s going to win the game?’ my best guess would be the 55% probability team. But, unless you explicitly ask it, it probably won’t tell you that it’s a 55% probability. If you try to suss out either the probability or the confidence that it has in a prediction, it’ll peel back the onion a bit further. Ultimately, what would be most useful for people to know is what those probabilities are. If I knew that it was a 90% probability, I'd probably bet very differently from if it were a 55% probability.”

Q: In light of the absence of Cinderella stories this year, how can brands pivot when those stories are not the tournament’s primary narrative?

Boyd: “Cinderella is part of the magic, and the irony is if there’s no Cinderella, then market the crown. From the standpoint of the men’s bracket, the power conferences hold the prominent brands, and there’s a certain predictability with them. The blue bloods will always have very loyal and devoted fan bases who buy a lot of merchandise. From a brand standpoint, we don’t have to be as worried about anchoring on a Cinderella that could disappear by the Elite 8, so instead we go with the crown.”

Q: How does the 'gambler's fallacy' manifest in March Madness? Do bettors tend to over-correct by betting on a 12-seed just because they saw a different 12-seed win earlier that day?

McCarthy: “It’s part of the human condition that we observe history and try to create some story around it to connect the dots. Oftentimes, the dots should be connected; oftentimes, they shouldn’t. You flip a coin, you flip it again, you flip it a third time; those are all independent draws. They're unrelated to one another. Even if you had 50 coin flips and they all came up heads, you were just exceedingly lucky. I think the same is true for these games. The men’s bracket has been really chalky, with many of the highly seeded teams winning. On the one side, you can create a story that maybe that trend will continue, or other people could say that lower-seeded teams that remain are due for an upset, and they’ll bet more on the underdog. This is an area where grounding yourself in sound analysis can be a suit of armor that protects from that fallacy.”

Q: Opendorse data shows NIL earnings spike 92% in March. Is this signaling that brands prefer "one-month wonders" who go viral in the tournament, or are they pivoting toward multi-year deals with underclassmen to build long-term equity?

Boyd: “I’d say both phenomena are happening. March has become a liquidity event. If I’m a player with the magic and my team goes deep in the tournament, I could really see my NIL take off because everyone wants to get on the bandwagon. If I’ve got freshmen and sophomores, I’m thinking there are future equities there. If I’ve got juniors and seniors, they’re the ones that are going to give me the return on investment. But we can have both. In the short term, you get the spike. In the long run, you can build on the return on investment from stars that have established themselves. We’ve got a lot to learn, and you can argue that we’re experimenting since it is the early days of NIL. We’re trying to figure out how best to take advantage of when these athletes shine, and March is the time to do it.”