Rewarding fan loyalty: designing a complex admin system in the age of AI

Alt title: Rewarding fan loyalty: designing a complex admin system in the age of AI?
Alt title: Designing a fan rewards system for professional sports teams in an AI driven world

The Problem

Every fan gets the same experience. That shouldn’t be true. Today, the Jump Admin has no way for a sports team to say: “If a fan does X, show them Y.” There’s no mechanism to recognize a fan’s behavior (like buying tickets, attending games, etc.) and respond to it with a different experience in the app.

Teams work around this manually. Someone pulls a spreadsheet each week of fans who bought tickets, then bulk-uploads it to grant them credits. Another team wants to show badges to fans who have attended a certain amount of games and currently has no way to do it at all.

The Goal in Plain Terms

Give sports team a tool to set up rules: “when a fan does this, put them in a group, give that group this experience.” Automate what’s currently manual. Make the fan app feel more personal.

The Complexity

This touches everything, except nothing existed yet.

The design challenge wasn't just a screen or a flow. It was figuring out how four separate parts of the product would need to connect in order for this to work at all:

Rules Engine: The backend that watches fan behavior and decides when a rule has been met.

Admin Tool: Where team staff configure rules – the screen no one had designed yet.

Content System: Where teams build the fan-facing app experience which currently can’t target individual fans.

Fan App: Where fans see badges, account credits, and other personalized content.

Additionally, engineers had different ideas about where things should live. The product brief hadn't been written. Three different client teams had three different needs. And the August 1st deadline was firm.

My Approach

Build shared understanding before designing anything.

Before opening a design tool, I needed to understand what exists today, what’s being built, what’s still unresolved, and where the team’s mental models diverge.

I used Claude Code to do this faster and more thoroughly. Not to generate ideas for me, but to help hold complexity, cross-reference sources, and surface questions worth asking.

I brought the judgement and AI brought the synthesis. AI compressed days of async research into a focused place pulling from Linear, Granola, Notion, and spec docs simultaneously.

Title

“Luxury” properties have high resolution imagery and beautifully curated interior design. “Non-lux” properties usually have lower quality photos, smaller aspect ratios, and interiors often aren’t as curated.
We created components that worked nicely with full bleed images and other storytelling components that provided space for imagery and text. Often times, guests staying at luxury properties are interested in hotel amenities and experiences, so we ensured componentry was available to showcase things like restaurants and spas.

🚧 Work in progress - to be continued 🚧