A government-funded study tested whether changes to restaurant ordering could reduce plate waste. At one Hong Kong restaurant, average weekday plate waste adjusted for sales revenue fell 41.9% between the baseline and intervention periods. Decisions Lab simulated the intervention and estimated a 40% reduction, a difference of 1.9 percentage points.
This comparison examines one measured outcome from one restaurant. It is a test of how closely the simulation matched that outcome, rather than a measure of accuracy across all decisions or settings.
The intervention and the simulation
Each person is modeled through a full day, and the restaurant meal is one part of that day. Activities, tasks, age, and body weight determine how hungry they are by the time they order, so someone who has been working since morning arrives hungrier than someone who ate earlier.
Staff in the simulation take the order. They can say how much rice is coming, offer a smaller portion at the same price, ask whether to leave off the fresh side vegetables, and allow cooked garnishes to be removed. A notice at the entrance and on the menu says a food-waste study is underway.
The researchers adjusted daily plate waste for sales revenue to account for changes in restaurant activity. The comparison below uses the reported weekday measure.
01
Plate waste
Edible leftovers, weighed each day after inedible scraps were removed.
02
Weighted
Each day's waste adjusted for that day's sales revenue.
The simulated reduction was 1.9 percentage points below the observed result
Reduction in weighted weekday plate waste
- Simulation40.0%
- Published pilot41.9%
A measured outcome gives the simulation a meaningful test
The simulation came within 1.9 percentage points of the reported weekday reduction. That gives us a concrete way to assess the model against a measured behavioral outcome: how much food diners left after the restaurant changed its ordering experience.
The simulation was configured independently, without knowledge of the published result. The field study compared a four-week baseline with a four-week pilot at one restaurant. It tested portion options, staff prompts, and notices together, so the result does not isolate any one change or establish accuracy in other settings.
Marketers can use Decisions Lab to test how customers may respond to a change before introducing it, whether that means revising a message, offer, or campaign. By comparing responses before and after the change, teams can identify what to investigate further and use real customer feedback to check the results.
Source for the field study: Lee, D., Wan, C., Leung, T. C. H., Rundle-Thiele, S., and Li, G. (2024). Application of marketing to reduce consumer food waste in restaurants. European Journal of Marketing. See also the government-funded final report (pages 31–32).