Skip to content
KPI Systems and Balanced Scorecard | Part 2 of 8

The Definition Sheet: One Page per Key Figure

Formula, scope, data source, collection method, rhythm, owner and known error sources on one page, with a dated version history.

Ask three people in a plant what on-time delivery was last month and you may get three numbers. One counts the ship date, one the agreed arrival date, one the date that was renegotiated with the customer last week. None of them is lying. The figure simply has no definition.

The definition sheet is the contract about what a number means. One page per figure, stored with the system rather than in a personal folder, and the first thing a new colleague reads.

What goes on the page

Field What goes in it
Name and purpose The figure and the decision it supports, in one sentence each
Formula Written out, with every term explained
Unit and decimals How it is reported, so that nobody rounds differently
Scope What is in and what is out
Data source System, report and field, named exactly
Collection method From a system, by hand or by sample
Rhythm Measured how often, shown how often, reviewed how often
Owner One person by name and role
Known error sources What makes the value wrong
Version and date Every change dated, with a one-line reason

For on-time delivery in the example plant, the sheet reads: shows whether we keep the dates we promised, and decides where scheduling capacity goes. Deliveries on the agreed date divided by deliveries due. Percent, one decimal. All customer deliveries of the plant, internal transfers and samples excluded. Shipping module, saved report “deliveries due”, field “agreed arrival date”. From a system, read weekly, reviewed monthly. Owner: logistics manager. Known error sources: rescheduled dates counted as agreed, ship date used instead of arrival date.

The definition that cost 2.6 points

The first value the plant reported was 90.1 percent. It was measured on the ship date, because that was the field the report happened to contain. The customers measured the arrival date. Recomputed on the agreed arrival date, the same month came out at 87.5 percent.

Nothing in the process had changed. The figure had not got worse, it had got true. And the improvement that followed over the next twelve months, to 95.4 percent, was real rather than a change of field.

That is the usual pattern. A plant that writes definition sheets for the first time will find that two or three of its headline figures were measuring something slightly different from what people believed. Finding that out is uncomfortable for a month and valuable for years.

Scope is where the arguments live

Formula disputes are rare. Scope disputes are constant. Do samples count as deliveries? Does a unit reworked at the station and passed on count against first pass yield?

Write both sides of the boundary, not only the inclusion. “All customer deliveries of the plant. Internal transfers and samples are out.” A reader who knows what is excluded can reproduce the number.

Changing a definition

Definitions change, because products, systems and customers change. Three rules keep the trend usable.

  • Date it. A change is a new version with a date and a one-line reason on the sheet.
  • Mark the break. The chart gets a marker at the period of the change, and comparisons across the break are labeled.
  • Recompute where you can. One year on the new definition keeps the trend readable, and the users hear about it before the next review, not in a footnote afterward.

One definition for the whole plant

Areas that report the same figure use the same definition. Where a local exception is unavoidable, it is written on the sheet rather than carried in somebody’s head. Otherwise the first question in every comparison will be whether the two numbers mean the same thing, and the meeting is gone.

The next part gives each defined figure the four numbers that make it readable: baseline, target, tolerance and reaction limit.

Newsletter

One Lean idea.
Every two weeks.