Annual reviews with a large customer tend to include a slide nobody in the room from the selling side has seen before. It carries a service level figure, and it is lower than the one reported internally, sometimes by ten points.

Of all the measures a large customer keeps on the businesses that supply it, service level is the one that governs the relationship. Quality issues are handled by exception. Price is negotiated once a year. Service is scored continuously, it is visible to the buyer every week, and it is the number that decides whether a range extension is granted or a promotional slot goes to someone else. Very little else in the commercial relationship carries that weight.

Which makes it worth knowing what the number actually says, and why buyer and seller are measuring the same deliveries so differently.

Being hard on yourself does not close the gap

I spent years inside an organisation that was unusually demanding about this. Case fill rate was the measure, calculated against the original order delivery date, and against every case on every line. It was the most exacting definition the systems could support, and it was chosen deliberately. We wanted to beat all our competitors on service. Nobody wanted a flattering number.

And still, for the key accounts where the customer’s own figure could be obtained, that figure came in consistently and disappointingly lower than our internal measure. Not because the internal measure was soft. Because the two were counting different events.

Where the two numbers separate

Five mechanics account for most of the difference.

• The baseline date. One side measures against the confirmed delivery date, after a planner rang the buyer’s office and agreed to move it by three days. The other measures against the date on the original purchase order. The call was real and the goodwill was real, but the buyer’s system never heard about it.
• Partial deliveries. A shipment goes out at eighty per cent, the balance follows two days later, and the order closes as fulfilled. On the other scorecard it is one line short on the original delivery, and depending on the rules the follow-up either counts as a second failure or does not count at all.
• The measurement point. Most manufacturers measure at despatch, because that is the moment they control. Large customers usually measure at receipt into their distribution centre. Everything in between falls on the manufacturing side: the carrier who missed the booking slot, the pallet four centimetres too tall, the missing label on an outer case.
• Granularity. A twenty line order with one line short reads as ninety-five per cent on one side and a failed delivery on the other.
• Refused product. Short on remaining shelf life, wrong pack configuration, damaged corner. The ERP records it as delivered. The scorecard records it as never having turned up.

None of these rules are unreasonable. A buyer managing thousands of relationships cannot maintain a bespoke definition for each one. But they compound, and they tend to compound in one direction.

Start closing the perception gap

Asking for the scorecard is the obvious start, and asking for the definition behind it rather than just the score. Buyers hand this over more readily than people expect, on the reasonable assumption that anyone who wants to know how they are measured intends to improve.

The harder step is rebuilding the internal measure to those rules and running it in parallel with the existing one for a quarter. The result is uncomfortable, but it is recognisable, and it usually points to one of the five mechanics rather than all of them. Frequently it is the informal date changes, which turn out to be cheap to correct once somebody accepts that a phone call is not a system transaction.

The gap is not evidence of poor delivery. It is evidence that two systems are counting different events.

Until the customer’s definition is intimately understood, we may be investing money and attention into maximising our internal metric while the failures the buyer actually records carry on unaddressed. Knowing precisely how a customer counts is what makes it possible to work on the things they value. It is the only reliable route to the top of their list.