Food-service waste: what the latest EU numbers mean for operators

The figures cannot tell an individual restaurant, hotel or contract caterer what its own waste is, what a reasonable level would be, or what a target should be. This follows from what an aggregate is.

An aggregate describes a population; a measurement describes a unit. The EU figure describes food waste across the EU, distributed across households, retail, manufacturing, food service and other sectors, and within each of those across an enormous range of operations with different formats, menus, guest bases and price points. The variation within the sector is far greater than any figure a single operator could derive from the aggregate.

The most common error is to treat the per-inhabitant figure as a per-person benchmark for a business, or the sector share as a target. Neither follows: a per-inhabitant figure divides national waste by the whole resident population, most of whom are not customers of any given food-service business, so using it as a per-cover benchmark compares figures with entirely different denominators.

There is a further distinction worth drawing, because this library discusses European food waste statistics in more than one place. A statistic describing the whole economy answers a question about national and European progress. A statistic describing the food-service sector specifically narrows the population but does not change what kind of figure it is. Neither is a site-level measurement, and neither substitutes for one.

The baseline an operator needs is built from its own operation, and the essential decisions are what to measure and what to divide by.

What to measure follows the same logic discussed elsewhere in this library: separating kitchen preparation, unserved display and plate waste, because each has a different cause. A single aggregate figure for the site cannot be acted on; the separated figures each point to a specific operational change.

What to divide by is the operator's choice and should match the business model. Covers served is the most common and suits a cover-driven operation; revenue suits one whose mix varies substantially across the period; meals produced suits a production kitchen. The denominator should be one the operation already records reliably, because a baseline built on an unreliable denominator will move for reasons that have nothing to do with waste.

Comparability across sites deserves caution. Two sites of the same business may have materially different baselines because of format, menu, guest base, opening hours and equipment, and treating a difference between them as a performance difference before establishing that they are alike is the same error discussed in relation to hotel benchmarking.

Food-service waste is seasonal, and a baseline measured in one period cannot be assumed to describe another.

Volume varies with the trading pattern — holiday periods, event calendars, weather and the local working pattern all move covers. Menu composition varies with the season, and different ingredients have different trim and yield. Guest mix varies too, and different guest types waste differently.

The implication is that a baseline should cover a period long enough to include the operation's range of conditions, or be built as a set of period-specific figures. A one-week measurement taken in a quiet period, extrapolated across a year, will misrepresent the operation and produce a target that is either unachievable or meaningless.

The measurement method itself also needs to be consistent. Waste measurement depends on weighing, and weighing practice — what is included, at what point, by whom, using what equipment — determines the figure as much as the waste does. The method should be documented and applied consistently, so that the time series compares like with like.

A baseline has done its job when it supports a decision. That means pairing the waste figure with a cost and a threshold.

The cost side converts the waste into money the operation is spending on food that is not sold. For the avoidable portion of the streams, that cost is the purchasing cost of the food thrown away, which is a figure the operation can obtain from its own purchasing records. The inedible portion does not carry that same avoidable purchasing cost, which is one reason the edible/inedible distinction matters.

The threshold is the level at which the operation would act. A threshold makes the measurement actionable: a stream that exceeds it triggers a specific response, and the response can be tested against the following period's figure. Without a threshold, the baseline is a number that is recorded and not used.

The decision that follows from a well-built baseline is usually specific and small: change the replenishment batch on one display station, adjust the portion specification on one menu line, review the ordering quantity on one ingredient. The European statistic tells the sector that the issue is material; the operator's own measurement tells the operator what to change on Monday.