The honest measure

Three numbers that are true and worth nothing

At the end of Bezirhan's second year, three figures were reported, and every one of them was accurate:

• 83.8% of the year's styles were built in the 3D system
• 214 virtual samples were made
• all four seats were in daily use

Every one of those is compatible with 468 physical sample rounds shipping and a net loss of USD 12,875.28. That is the test of a bad measure. It can be excellent while the thing it exists to describe is not happening.

There is a sharper way to say it. Adoption measures your own compliance with your own instruction. You told people to use the system, and they did. That is worth knowing on a project plan, and it is not a result, because nothing outside the building has changed.

Two numbers are results, and they are the only two this course will accept: sample rounds removed on real styles, and days saved on real critical paths. A third, days banked, decides whether the saved days meant anything.

First, the ceiling: a calendar can only give up what it contains

Before promising anything, divide. Bezirhan's development calendar for a tailored jacket, from tech pack to bulk fabric committed, is 74 days. The sampling activities sitting on the critical path are the proto round at 11 days, the fit round at 11 days, and 4 days of internal review. That is 26 days, or 35.1% of the calendar.

That is the absolute ceiling: what you would save if sampling took no time at all, which no process achieves. The real ceiling is lower, because a round's elapsed days are not all its own. Four of the 11 overlap with fabric ordering, which continues whatever the sample is doing. Recoverable: 7 plus 7 plus 2, so 16 days, or 21.6%.

A programme promising to halve the development calendar is promising something the calendar does not contain. The division takes two minutes, and it should be on the first page of the business case. It is the number that keeps a good programme alive when somebody asks in month nine why the calendar is not half of what it was.

What a removed round is actually worth in days

Fourteen fit rounds came off in the pilot. The tempting arithmetic is 14 rounds at 11 days: 154 days.

The honest arithmetic asks a question about each style. Was that round on this style's critical path? For 9 of the 14 it was. For the other 5 the style was waiting on a nominated trim, a colour approval or a buyer's own comments, and removing a fit round moved the ship date by nothing at all.

• 9 styles at 7 recoverable days = 63 days
• 5 styles at 0 days = 0
• Across all 14: 4.5 days a style

The headline figure overstates it by 2.44 times. Not because anybody lied. Because 154 is what you get when you multiply two true numbers that should never have been multiplied.

Days saved and days banked are different columns

A saved day is worth something only if it bought something. Of Bezirhan's 63 days, 21 were converted into an earlier bulk fabric booking at Ravenglass Mills. That is three styles at 7 days each, and it took those three styles off an air-freight uplift worth USD 4,180. The other 42 were absorbed: the style arrived early at the next gate and waited there.

That is not a failure, and it should not be hidden. It is the ordinary behaviour of a calendar with more than one constraint in it. And it leads to a rule worth writing on the ledger itself. Count days saved and days banked in separate columns, and never add the second to the first. A programme that reports only saved days is reporting a number nobody in the finance office can find.

The ledger, and why publishing a negative number is what saved the programme

The ledger is one row per style and three columns: rounds removed, days saved, days banked. A fourth free-text column says what the banked days bought. It is kept by the merchandiser who runs the style, not by the programme.

Bezirhan put the table above in front of its owner at the end of year two, negative number included. That is why the retirement rule of lesson 5 was approved, and why the programme reached the year-three figure of plus USD 12,568.97 instead of being cancelled.

A programme that reports adoption cannot be corrected, because nothing it reports can go down. A programme that reports rounds and days can be argued with, and being argued with is the only route to a rule that changes anything.

What to take away

1. Adoption measures your compliance with your own instruction. It can be excellent while nothing has changed.
2. Divide before you promise: Bezirhan's sampling occupies 35.1% of the calendar and can recover 21.6% of it, and no programme beats what the calendar contains.
3. A removed round saves days only where it was on the critical path. Bezirhan's was in 9 cases out of 14.
4. Multiplying rounds by elapsed days overstated the saving by 2.44 times, with no dishonesty anywhere.
5. Days saved and days banked are different columns. Only the banked ones bought anything.
6. Publish the honest number even when it is negative. It is the only kind of number a rule ever gets written from.