Off-Season

At ESPN BET, engagement predictably craters every summer as football and basketball wrap up.

In 2025 that pattern held true: monthly active users stayed strong into the NBA and NHL playoffs, then dropped sharply within a couple of months as those seasons ended, per Amplitude. That's not a UX nuisance, it's a direct hit to year-round revenue and retention.

Earlier segmentation analysis had already surfaced a signal worth chasing: a meaningful share of users showed a latent willingness to bet on other sports, they just weren't acting on it. Over 4 months, working alongside product design, product management, and market research at theScore, my job was to figure out why that willingness wasn't converting into behavior, and build a strategy the team could actually act on.

The two clusters showed meaningfully different betting patterns going in, a gap wide enough to justify analyzing them separately.

I started with what the data could tell me on its own: Amplitude behavioral analytics. To get the why behind it, I designed a survey sent to roughly 29,000 users active over the past year, split across two behavioral clusters Amplitude had already identified: casual, single-sport bettors with high hold but lower overall engagement, and multi-sport bettors already betting across several sports every week, with the heaviest use of live betting.

Averaging them would have produced one muddier, less actionable recommendation. Before committing the field time, I brought the two-cluster plan back to market research for a second read: a check on whether the split would actually get us to the goals we'd aligned on, not just a plan that looked reasonable on paper.

330 people responded, about 1% of the list, with a reduced base on some questions. Everything below is directional rather than precise, so I read the size of the gaps between answers rather than the exact shares.

Aligning before designing anything

Before drafting a single survey question, I got product design, product management, and market research in a room to align on goals. That meant making sure this work wouldn't run parallel to something already underway elsewhere in the org, and making sure whatever came out of it would be genuinely useful to each of them, not just directionally interesting to me.

The data reframed the whole problem: users weren't uninterested in other sports.

Nearly half said they bet just as often in the off-season as during it, and almost the same share stayed at least somewhat active. Only a small handful stopped betting altogether. Most had already placed a bet on a new sport in the past year, and many had tried to learn about a new sport specifically through betting on it.

What stopped the rest wasn't apathy, it was confidence. The top reasons users gave for hesitating to bet on an unfamiliar sport clustered around a lack of knowledge, not a lack of interest: confidence in how to bet smart and familiarity with the rules dominated the list, with wanting more incentives a distant second. A general lack of interest in the sport itself, the one reason with nothing to do with confidence, ranked lowest of all.

That reframing pointed at promotions first. A strong majority of respondents said risk-free or promo bets would make them feel more confident trying a new sport, and no-cost trial offers were by far the most-requested type, so promotions became the obvious lever to drive that first, low-risk try.

Promotions alone would have been the easier pitch, and I argued against it.

They only line up cleanly with a handful of marquee events each summer, and plenty of promo-driven users disappeared again as soon as the offer did, visible in the same Amplitude monthly-active view. A promo also doesn't teach anyone to read odds or understand a format they've never seen.

Most users said in-app promotions and banners were where they discovered new sports in the first place, which made the app itself the natural place to pair the offer with the content. It also happened to be a lever unique to ESPN, built on its media and content strength in a way no sportsbook competitor's promotion could match.

It mattered for how the recommendation would land, too. I already expected marketing to be unreceptive to a plan built on promotions alone, and education gave the strategy a second leg that didn't depend on discounting.

I brought the recommendation to product management, design, and marketing as a deck built around a story, not a list of findings.

The story: people did have an appetite for off-season betting, they just didn't want to risk money on a sport they barely understood.

To back that up rather than just assert it, I showed our own product data confirming users were already treating betting itself as a way to explore an unfamiliar sport, and that the app was already their main point of discovery for those sports in the first place. The behavior we wanted to design for was already happening at a small scale, not hypothetical.

I also broke out which sports were already gaining interest fastest, so the room had a concrete starting point for where the education effort should focus first instead of trying to cover every sport at once.

Closing on the two-part structure, promotions for the short-term trial and education for the long-term habit, gave stakeholders a strategy they could act on immediately and one they could keep building against afterward.

The survey went out during the summer, close to a major tennis tournament, and looking back, that timing could have skewed the read.

A high-profile event like that pulls casual sports attention and betting activity in ways an ordinary summer week doesn't, so some of what looked like steady off-season engagement may have been the tournament itself, not evidence that users stay engaged through a quiet off-season generally. If I ran this again, I'd time the fielding to avoid overlapping with a major event, or at least account for it in the analysis, so the confidence-gap finding wasn't riding on a moment that wasn't representative.

I'd also want more than 330 responses behind a recommendation this size. The gaps between the answers were wide enough to act on, but a 1% response rate leaves room for the people who did reply to be the ones who already cared most about betting.