Hyperlocal campus merchandising moat
David Lin, CEO of Duffl, on the economics of hyperlocal ultrafast delivery
The real moat here is not faster delivery, it is turning each campus store into a live merchandising system that learns local taste faster than a national chain can. A Berkeley store can carry different drinks, protein bars, fruit, and private label items than an Arizona State store because Duffl sees every search, click, add to cart, repeat purchase, and delivery address, then uses that data to generate weekly buy lists for each location.
-
Duffl already ties merchandising to hard behavioral signals. The company tracks views, clicks, add to cart, remove from cart, conversion, repeat purchase timing, search wording, and address level patterns, then pushes supplier specific purchase recommendations to store managers and student operators.
-
This local curation matters because campus demand is unusually concentrated and culturally specific. Duffl says students order four to six times a week, mature stores are profitable, and 80% of customers discover the service through social connections, which makes assortment part of campus identity, not just inventory planning.
-
The closest analogue is less DoorDash and more a mix of Gopuff and Instacart. Gopuff has used direct consumer feedback to build private label products, while Instacart has built personalized recommendations and ads on top of search and purchase data. Duffl is applying that same data loop at a smaller, campus by campus level.
If Duffl keeps improving this SKU test loop, each new store becomes easier to launch and harder to copy. The next step is a recommendation layer that does for snacks what TikTok does for videos, while giving brands a reason to pay for sampling, placement, and eventually ads inside a campus specific storefront.