Measuring On-Time Delivery Properly: The Definition Choices That Change the Number

On-time delivery is one of the most watched numbers in manufacturing and distribution, and one of the least consistently defined. Two companies, or two departments in the same company, can report very different percentages for the same shipments, and both can be right under their own rules. The choices below change the number more than most people expect.

Which date are you measuring against?

There are at least three. The date the customer asked for, the date you promised, and the original promise before it was revised. Measuring against the revised promise flatters you, because every late order can be re-promised until it is on time. Measuring against the customer's requested date tells you how well you meet demand. Many companies report both and learn a lot from the gap.

What counts as on time?

Is it the exact day? A window, such as a day early to a day late? Does shipping early count as on time, or as a miss because the customer cannot receive it? Different customers care about different things, and a window that suits one is wrong for another.

Order level or line level?

If an order has ten lines and nine ship on time, was the order on time? At order level it is a miss. At line level it is ninety percent on time. Line-level measurement is usually more informative and harder to hide behind, but it needs to be stated clearly.

What about partial shipments?

A shipment that arrives complete but split across two deliveries is different from one that arrives short. Decide how partials count, and whether the original due date applies to the remainder.

Ship date or delivery date?

If you measure the day the truck leaves, you are measuring your own performance. If you measure the day goods arrive, you are measuring the customer's experience, including the carrier. Both are valid. They answer different questions, and mixing them causes confusion.

What is excluded?

Delays caused by the customer, changes they requested after ordering, and backorders that were agreed in advance are sometimes excluded. Exclusions are legitimate where they are rule-based and documented, and corrosive where they are applied case by case.

Which period does a delivery belong to?

An order due in March that ships in April belongs to March if you measure by due date, and April if you measure by ship date. Due-date-based reporting is fairer and harder to game.

What we recommend

Report on-time to the customer's requested date and on-time to the promised date, side by side. Measure at line level, with a clear partial-shipment rule. Attribute each delivery to the period it was due. Document the definition on the dashboard itself, so anyone reading the number knows what it means.

Why it matters

A number that people do not trust changes nothing, and a number that can be gamed changes the wrong things. A clear definition lets operations, sales and customers talk about the same thing. It also makes trends meaningful, because you know the rules did not shift underneath them. If the number is shown on the floor, design the display for the people who can act on it, as covered in how to build a production dashboard your operators will actually use.

For the wider framework of agreeing what to measure, see building a KPI framework your whole organization agrees on and the 8 KPIs every manufacturing leader should be tracking. The same definitions issue appears when scoring suppliers.

If your on-time delivery number is argued about more than it is acted on, reach out. We usually start by writing down how it is calculated today.