The Pitfalls of Averages

By Kevin Roberts, Project Engineering Manager

“I have a good handle on my operator productivity, but some days we just can’t get everything out the door, and other days I send people home early”. Is this your frustration? It’s possible that the answer you’re searching for is hidden in the numbers.

Perhaps you are confronting the disparity between averages and actuals. Even with good summary data on average lines picked per order and average productivity per operator, designing to actual detail order configurations can have a huge effect on operator productivity.

Consider the orders below:

Both order groupings average 5 lines per order. Orders A-E are all similar, varying less than 20% from the average. Orders V-Z however are from 1/5th to 3.4 times the average. The effect of these two order groups on picking labor can be quite different. If your system is designed for picks to be placed into a tote sized to handle 4-6 line items, what happens when there are 17 lines? Or one? Is your picker making extra trips to get additional totes and registering new orders? Can your current operation really accommodate fluctuations? Can you handle your actual orders not just an average order?

It may be time to take a closer look at your orders. While some picking operations may not deviate much from an average order size, many vary greatly. These variations can be caused by seasonal or customer ordering patterns or sudden popularity of specific products. Reviewing individual orders and then designing your operation to accommodate the actual order types will stabilize your ability to respond.

An effective way to look at your orders is to categorize them by the number of lines per order. Consider a recent client who was struggling with decreasing picker productivity and an unstable daily labor requirement. By analyzing actual individual order data collected from a recent peak month’s activity, it became clear that the existing systems (both physical picking methods and order release techniques) were no longer tailored to recent order profiles. The average of 16.2 lines per order was not far from the original system design. However, the review highlighted a large increase in the number of single line orders.

Once realized, picking methods were adjusted to reflect the current order mixture, which returned picking efficiency to its previous levels.

This illustrates a typical situation in which looking at average order summaries does not adequately reflect the situation. A review of actual order data is required for an accurate picture.

Detailed review of specific orders, line by line, should not be restricted to just the number of lines. Sometimes changes in the cube of the products can cause an existing system to become inefficient. Occasionally, the review must be extended to the number of pieces per line.

In any case, a closer look at your actual order composition will provide you with a more accurate assessment of your operation when determining where productivity and efficiency improvements can be made.