Why a hotel’s guest count is not a breakfast order

David Jiang, a chef with 32 years in kitchens, had a planning number that looked useful: how many guests had breakfast included. It was also misleading. In Winnow’s account of his work at Aloft Shanghai Hongqiao, the executive chef explains that up to one in ten of those guests did not come to the buffet.[1]

A full hotel can still leave a particular serving dish untouched. The kitchen must decide how to distribute ingredients, preparation time and the risk of running out across the menu. An occupancy forecast gets it only part of the way.[2]

The hotel introduced Winnow Foresight in June 2026. Winnow’s 5 October account reports a 67% fall in breakfast food waste, measured in grams per cover, during the first two months. That means less waste per diner. This supplier-published result does not isolate the software’s contribution from the team’s other changes.[1]

The production plan starts before service

Launched on 29 September after development with Hilton chefs, Foresight targets hotels with relatively small food operations. It is a mobile web app, available on its own or alongside Winnow Vision, the camera-and-scale system used to measure discarded food. Existing phones or tablets can access Foresight; kitchens need not install a Vision unit to use it.[3]

That distinction matters commercially. A prediction tool tied to a new piece of kitchen equipment has to justify both purchases. Separating the planning service from dedicated measuring hardware lowers one barrier to trying it. The practical question becomes whether the routine is useful enough to survive a busy breakfast shift.[3]

According to Winnow’s development account, the plan combines occupancy with the hotel’s own record of what was served and left over. It recommends quantities for individual items, and chefs can adjust them. After service, the Snap & Go function estimates leftover weight from a photograph; the result feeds subsequent forecasts. The system therefore has two jobs: turn a guest forecast into a menu, then learn from what remains.[2]

A useful way to judge that feedback is to ask what it changes tomorrow. “Less waste” is an outcome; “make a smaller first batch of this dish” is a decision. A daily system earns its place when it makes that second instruction easier to act on, while leaving the chef able to account for an unusual service.[3]

One buffet, different directions

Winnow says the product was developed with Hilton teams over three years, including kitchen visits in nine countries. As of its 6 October account, it was live in 13 Hilton hotels across 11 countries. Those Hilton deployments are separate from Jiang’s Shanghai case.[2]

For an operator considering such a system, the relevant comparison is a matched breakfast service: waste per diner, food cost per diner, and whether guests still find the dishes they want. A lower waste total on a quieter morning answers less than those measures together. Recording shortages alongside leftovers would also reveal when a leaner plan has become too lean.

At Aloft, the reported changes were specific: leftover fried noodles fell 83%, while the data also identified white-fleshed dragon fruit as a popular choice worth keeping sufficiently stocked. The same buffet needed less of one thing and enough of another.[1]

Report an error or suggest a correction