Fulfillment Tycoon
Run an outdoor retailer’s fulfillment for a year: twelve turns, ~9,000 orders a month, priced by a real sourcing cost model. Same seed for everyone.
The Challenge
Fulfillment Tycoon is a strategy game built on a simulation engine, not a spreadsheet of made-up numbers. Twelve monthly turns running Cairn Outfitters, a fictional outdoor-gear retailer: route orders, set buying postures, open DCs, survive vendor outages, a viral demand spike, carrier rate increases, and Cyber Week. The challenge was twofold — make it genuinely fun to play, and make every number on screen defensible: each order priced through zone-based rate cards, cartonization, and a seven-component cost model, with a global leaderboard where every score is verified by replaying the player's full decision log server-side.
My Approach
I built the engine from a frozen contract spec with a formal raise→review→amend loop — 23 documented questions where implementation pushed back on the spec. The most important design changes came from practitioner knowledge catching where the model diverged from how retail actually works: split shipments were repriced from a token annoyance penalty to the real cost of shipping a second package; clearance markdowns were added when the model's profit-optimal inventory position held far more cover than any merchant would run; seasonal demand curves were rebuilt so Thanksgiving through Cyber Monday carries real weight. Same-seed determinism runs through everything — a game is its decision log, replays are byte-identical, and common random numbers mean the what-if panel can honestly say the difference is policy, not luck.
Impact
Technologies
What's Next
- →Publish the full case study — what the model got wrong and how I knew
- →MILP-based optimal-routing benchmark to replace the current best-shipped-policy comparison
- →Playtest rounds and daily-seed leaderboard seasons