A customer receives an order and one item is missing. Or the box is damaged. They phone, the company apologises, sends the missing part or refunds the money. The case is closed and the next order is waiting.

By the end of the week the team has closed several cases like this. It has done a lot of work, but most of it was repair. The error was made days earlier, at a table, on a shelf or during loading, and nobody saw it.

This happens in warehouses, in deliveries, in production and anywhere work passes from one pair of hands to another. The details differ, but the pattern is the same: the check was placed where the error is most expensive to find.

Note where the error was found

Take the last ten errors and write two things for each: where it happened and where it was found. Often the two answers are far apart. An item was picked wrongly and found by the customer. Weak packing was done at the table and seen only after transport.

The distance between those two places is the cost. Every step an order passes unseen adds waiting, returns and people dealing with it.

An error costs less where it happens than where it is found.

Do not do this to find out who made it. Do it to see where the last step was at which the error could have been stopped cheaply.

Move the check closer to the source

A check at the end of the process is not wrong. But it catches only part of the errors, and when it finds one, the work around it is done. A check closer to the source is cheaper, because it stops the error before more work is built on top.

Some practical ways:

  • Confirm at the place of work. When an item is picked, a scan or a simple mark shows at once whether it is the right one, not after the order is closed.
  • Let the next step check the one before. The person packing spends two seconds seeing that the list matches what is in front of them.
  • Make the error hard to make. A clear label, a fixed place or a physical divider that does not let the wrong item in.

The useful question is not who made the error. It is the first step where we could have seen it.

Start with the error that repeats most often and move its check one step closer. Do not change everything at once.

Close the loop with the cause

A single error calls for a correction. An error that repeats across different people calls for a change in the process. When an error returns, ask why it happened, and ask again, until you reach something you can change. The Five Whys tool gives a simple structure for this.

Record damage where it happens, with the step and the cause. The Damage Control tool does this without much extra work. After a few weeks you will see clearly where errors collect.

A good check tells you where to look. It should not tell you whom to blame.