I have been a big advocate of using Perfect Order metrics to measure how well supply chains are able to meet customer demand. To get credit for fulfillment under a Perfect Order measurement system, everything about that order needs to match exactly how the customer expects it. This includes whether the order was delivered having the right products, right quantities, right place, right delivery time, right product quality, and right invoicing. It is a great measure of how well a supply chain is doing in matching supply and demand. But it does have a downside: it's a measure that is exclusively from the customer's perspective. It does not measure how efficiently supply and demand are being matched from a business's perspective. It does not address a business question such as: Are we running through hoops and losing our shirts too often in response to customer demand? Or are we doing order fulfillment optimally? For these reasons, Efficient Perfect Order metrics are better.

Efficiency is an issue because many supply chain personnel love to play the role of “Everyday Hero.” This person routinely solves problems to satisfy customers. What could be wrong with that? The customers are happy and the employees at the end of the day feel great. It's a win-win for all concerned, except perhaps for the company itself. The everyday problems get masked and extra costs are incurred from expediting and processing orders in non-standard, more costly ways.

I realized that “Everyday Heroes” could be an issue during a presentation at a conference that I attended several years ago. Two warehouse managers from a pet foods company were boasting that throughout their company's long history no order was ever shipped late — 100 percent on-time shipments. Amazing!

They went on to tell a story about a time an order could not be shipped on time because they were out of stock on one of the items ordered. Rather than ship it late, one of them went into a retail store and bought the item there. They added the purchased item to the shipment and sent the order out on time. What's wrong with this picture? After all, they kept the company's perfect shipment record, the customer was happy, and they solved a problem. However, there are a few concerns.

First, they bought the product at the retail price and sold it for less — so the company lost money on the shipment itself, without even counting the extra order processing costs. More importantly, they solved a problem without documenting that there ever was a problem. When this happens, the problem will likely recur. I won't even address whether it is legal for a company to buy its own products at a retail outlet—after all, one could inflate sales figures and stock prices that way.

Generally, supply chain planners benefit from understanding when things do not go according to plan. For this reason, exception management events need to be made visible to them to help prevent routinely occurring problems.

Perfect Order metrics are not perfect because they do not measure what happened when “Everyday Heroes” processed orders in non-standard ways to meet customer expectations in order to get credit for a perfect order. For example, did they take extra steps to correct mistakes or expedite an order to make up for a delay? Essentially, exceptions are not documented using only Perfect Order metrics, which makes it impossible for planners to do a root-cause analysis of the recurring problems that might be corrected over time.

While supply chain plans are never 100 percent accurate, planning is necessary to reduce the number of exceptions. Do little to no planning and most customer transactions become exceptions. Do a ton of planning and you'll still get exceptions because of the uncertainties in demand and supply. So there is a limit to how much planning should be done. That said, a cardinal rule is that planning processes should learn from mistakes in order to improve. Implementing an Efficient Perfect Order measurement system is one way to do this.

Under this system, Perfect Order metrics are tracked in accordance with general principles (see Exhibit 1). However, added to these metrics are those that measure how often and what happened when orders were not processed using less-costly standard processes. These include having to manually process an on-line order, touch an order more than planned, perform duplicate steps, and expedite to make up for delays, as well as not getting paid on-time. This additional set of metrics is useful for doing root-cause analysis of the service glitches that were prevented and would not be detected by Perfect Order metrics.

The main long-run benefits of implementing an Efficient Perfect Order rather than just a Perfect Order measurement system are reduced order management costs and improved customer service. Of course, the system will never get rid of all “Everyday Heroes” because they would still be needed to handle the exceptions that will crop up from plans that cannot fully account for the uncertainties in supply and demand.