How we count

Pounds

Every log is a weight read off a scale and typed in by the crew before the food goes in the trash. If the food is still in a pan, the crew picks the pan and the app takes off its empty weight, which the kitchen sets by weighing its own pans. We never estimate pounds from photos or sales.

One exception is food a kitchen chooses to count, like muffins or milk cartons. The owner weighs one piece, and the crew counts pieces. Those logs are marked counted in the open data.

The other is the pan sweep at the end of an event. The crew taps how full each pan is (full, 3/4, half, or a quarter), and the app multiplies that by the weight of a full pan of the dish, which the kitchen weighs once. Those logs are marked estimated in the crew app and the open data, and they lower the data quality score.

Shifts

A kitchen sets how many shifts it runs, up to four, and when each one starts. Every log is tagged with the shift it was saved in, by the kitchen's own clock. A shift is not missed until it ends.

Coverage

Coverage is the share of expected logs that came in, counted one shift at a time. If a shift expects 5 and gets 7, it counts as 5 of 5. Extra logs in one shift never cover for a missed one. A missed shift is a miss, never a zero.

The 80% line

When coverage is under 80%, the number is greyed out everywhere it shows. We still show it, because hiding it would hide the misses too.

Meals

Pounds divided by 1.2, Feeding America's figure for the weight of one meal. It's a way to picture the amount, not a claim that the food was still good to eat.

Baseline and change

The first 14 calendar days from a kitchen's first log are the baseline, and they are weigh-only. From day 22, the kitchen's page compares the average pounds per logged day over the last 7 days with the baseline average. One kitchen with no control group is a case study, not proof, and we say so on the page.

Thrown out and rescued

Each log says where the food went. Trash, compost, and down the drain count as thrown out. Donated, staff meal, sent home, and animal feed count as rescued, the way ReFED counts surplus food. The huddle and the baseline use both together, because the goal is not to make the extra food at all.

Dollars

Pounds times the cost per pound the owner enters for each item, from invoices, or the kitchen's single cost per pound when an item has none. The value is saved with each log, so a price change later doesn't rewrite the past. Food cost stays private and never goes on a public page.

Climate estimate

ReFED puts U.S. foodservice surplus at 12.5 million tons in 2024, linked to 63.6 million metric tons of CO2e from growing the food through throwing it out. That works out to about 2.5 kg of CO2e per pound. We apply it to thrown-out pounds only. It's an average for the whole industry, not a measurement of your kitchen. Source

Served counts

Kitchens can enter how many transactions, meals, or guests they served each day. The crew screen uses it to show pounds per 100 transactions or per meal served. Those counts stay private. They never go on a public page or in the dataset.

Cook-to number

For a recurring meal, the suggestion is the average served over the last four times that meal was served, leaving out days marked as a holiday or other event when there are enough plain ones. Once a kitchen has two past meals with both reservations and a served count, it uses reservations times the kitchen's own show-up rate instead. A marked event, like a holiday or bad weather, changes the number only by how much it changed the kitchen's own past meals. Then it adds 5%, which the kitchen can change, and suggests one backup dish. Forecasts from past days missed real attendance by 2 to 3% in a study of 21 school kitchens, against 20 to 40% for cooking for everyone enrolled (Malefors 2021).

Tray scores

The crew scores each serving left on a tray as none, a quarter, half, three quarters, or all of it. The share left is the average of those scores. Food a person saved for later counts as eaten. Scoring by eye this way agreed closely with weighing for 87% of foods in a school study (Getts 2017). Tray scores stay in the crew app and never go public.

Diversion rate

At an event, the crew counts bags as they leave. The diversion rate is compost and recycling bags divided by all bags. It counts bags, not pounds, so it's a rough guide.

What we never publish

Notes, photos, sales, served counts, anything from the kitchen tools, and anything about a person. A kitchen stays anonymous unless it chooses a public name, and it is only public if it turns publishing on.