Demos
Two browser toys for the problems I work on: how ranking changes consumer surplus under costly search, and how a brand promotion’s “lift” splits into incrementality versus cannibalization. Both are stylized teaching models, not production estimates.
Search-ranking welfare
Shoppers search an online travel agency from the top of the list downward. Inspecting a hotel reveals a match value; each inspection costs time. They buy at most one room, with perfect recall, and stop when the expected gain from another click is below the search cost. Because people follow the list, the platform’s ranking changes who gets found—and therefore surplus.
This is a pedagogical sequential-search model in the spirit of my OTA ranking paper. The paper’s empirical estimates ($8.11 from random → proprietary; −$23.87 from dropping the top five) are not reproduced here.
Purchase share by hotel
Current list
Incrementality vs. cannibalization
A marketplace boosts one brand on a new-brand surface. Dashboards often treat every extra order as incremental. In reality, some of those orders would have gone to substitute brands anyway. This toy maps promotion intensity into gross attributed lift, then splits it into net-new GMV versus cannibalized GMV—within category and across category.