Reducing waste with better forecasting, product by product

5 days ago
3 min read

Introduction

Waste is the purest form of loss. Unlike a discount, it buys nothing, and unlike an overhead, it produces no capability. Whatever is thrown away was paid for at full cost, and the margin required to replace it is several times its value.

Yet in most businesses waste is recorded as a single number in the accounts, if at all, and managed by exhortation. The reason is that measuring it properly is tedious: it means recording what was discarded, by item, with a reason, every day. That is exactly the kind of data collection that becomes worthwhile once something can analyse it, because the pattern is where the money is.

1. Reducing waste with better forecasting begins with measuring the waste

You cannot forecast against an unknown.

Record what is discarded, by item, quantity and reason, daily. Two weeks of this data tells you more than a year of aggregate figures, and the concentration is always surprising: a few lines usually account for most of the loss.

2. Separate the reasons, because they need different fixes

Waste is not one problem.

Over-production, expiry, damage, spoilage, preparation trim, customer returns, order errors. Forecasting fixes over-production and expiry. It does nothing for damage or handling, which need process changes instead.

3. Forecast at item level for the items that matter

Selective effort.

Forecasting everything is unsustainable in a small business. Forecast the twenty per cent of lines that produce most of the waste, and manage the rest with a simple rule. This is where the return concentrates.

4. Use the shortest useful time bucket

Daily, in perishable businesses.

A weekly forecast is useless for something with a three-day life. Match the forecast interval to the shelf life or the production cycle, or the forecast cannot inform the decision it exists to inform.

5. Include the local factors that actually drive demand

They are frequently dominant.

Weather, day of week, school terms, paydays, local events, nearby closures. In food, retail and hospitality these often explain more variation than any trend, and they are the reason a generic forecast disappoints.

6. Decide the cost of a shortage against the cost of waste

The trade you are actually making.

Zero waste means running out, and running out costs a sale and sometimes a customer. Establish which is more expensive for each line, then set the target service level deliberately rather than aiming vaguely at less waste.

7. Reduce batch sizes where you can

Frequently more effective than better prediction.

Producing twice a day in smaller quantities cuts waste more reliably than forecasting once a day accurately. Where the setup cost permits it, this is the more robust answer.

8. Give the waste figure to the people who create it

Visibility works.

A daily or weekly figure, by line, shown to the team who produce and handle the items, changes behaviour without any policy. Waste managed centrally and reported monthly changes nothing.

9. Recheck the forecast against actual waste, not against sales

The right feedback loop.

Sales accuracy and waste are related but not identical, because you can hit the sales forecast and still discard a great deal through mis-timing within the day. Measure both.

Look at what happens to the waste before you claim a saving. Waste eliminated by running short is a transfer to lost sales, not a gain, and it will show up in the revenue line instead.

Conclusion

Measure what you discard by item and reason before trying to forecast it away.

Separate the causes because forecasting only addresses over-production and expiry, concentrate the forecasting effort on the small number of lines producing most of the loss, match the forecast interval to shelf life, include weather, day of week and local events which frequently dominate, decide deliberately whether a shortage or the waste costs you more, reduce batch sizes where setup cost allows because it is more robust than prediction, show the figure to the people handling the items, and measure the result against waste and revenue together.