Picking accuracy

How picking accuracy is worked out

The sum is the number of lines picked correctly, divided by the total number of lines picked, multiplied by one hundred. The arithmetic is the easy half. The hard half is how you find out that a line was wrong, because that decides what the figure is actually measuring.

An error can surface in three places: at the packing bench, where a check weight or a second scan catches it; in your own goods-in, when a customer sends the wrong item back; or in a complaint, where you hear about it and the item is never returned. Most warehouses only ever see the first two. Write down which sources you count, and expect the figure to get worse the first time you start counting a new one.

There is no good number

A distributor of fasteners with four hundred near-identical parts in adjacent bins is working against the physics of the job. A book wholesaler, where every item carries a unique barcode and looks nothing like its neighbour, is not. Both can run a tight operation, and their figures will not match.

So the comparison worth making is against yourself: this quarter against last, and one zone against another in the same building. A single zone drifting while the rest holds steady is a far more useful signal than the site average.

Read it next to picks per hour. Accuracy and speed trade against each other, and a figure that improves while the other falls apart has not improved anything.

In practice

Of those 10,000 lines picked in the week, 40 were wrong. 9,960 divided by 10,000 is 0.996, so picking accuracy was 99.6 per cent.

Now split the 40. Twenty-eight were caught at the packing bench and corrected before the parcel left. Twelve reached the customer. Measured at the pick face the figure is 99.6 per cent; measured as the customer experienced it, 9,988 of 10,000 lines were right, or 99.88 per cent. Both are true and they differ by a factor of three, so a report that does not say which one it means is not telling you anything. These numbers are an illustration.

What moves the figure

• Scanning at the pick face. Confirming each line against a barcode catches the wrong-item error before it travels. Picking from a paper list and confirming the order in one go at the end does not.
• Where look-alikes sit. Two sizes of the same part in neighbouring locations will be picked wrong eventually. Separating them is cheaper than training around them.
• Labelling and light. A location label a picker has to crouch to read, or a dim aisle at the end of the rack, produce errors that look like carelessness.
• Split batches in one location. Two batch numbers in the same bin makes the correct pick a matter of which box the picker reaches into first. See stock accuracy.
• Pressure at the end of the day. The last hour before the carrier cut-off is where most of the week's errors are made. A wave plan that finishes early is an accuracy measure.
• New and temporary staff. Peak brings in people who do not know the building. Errors cluster around them, and the fix is in how they are paired and what they are allowed to pick, not in the figure.