A bonus system nobody ever triggers is not demanding, it is broken

The instinctive reading of that table flatters head office: standards are high and the teams are not there yet. The correct reading is the opposite. A system nobody ever reaches has stopped measuring performance and is measuring its own construction.

Three scales coexisted inside the same company

Going back through the file, I found three different weightings in circulation at the same time.

The one displayed in the dashboard. The one in the guide handed to managers a few months earlier, with its own weights and a threshold of 90. And the version actually applied by the generator, with different weights again and a threshold of 80.

The document the managers were holding therefore announced a rule that was no longer the one being calculated. They were working conscientiously towards a target that no longer existed.

And the real score generator was sitting in a folder marked obsolete, excluded from searches by a filter. That is an archiving detail, and on its own it explained half the confusion.

The aggregation bug, and what it hid

The old method averaged weekly rates, each capped at one hundred per cent. That cap is the error.

An outlet that had declared only 29 % of its services scored 9.6 out of 15 on that criterion. The monthly method, which compares delivered against due over the whole period, gives it 4.3 out of 15.

Averaging capped ratios rewards regularity of reporting, not volume reported. An excellent week never compensates an empty one, but an empty week drowns in the average. So the system was scoring attendance and believed it was scoring output.

The forgotten exemption

Two outlets out of four were being scored on a base of 90 instead of 100, under an exemption granted several months earlier for a reason that had ceased to apply.

The data that would have allowed that exemption to be retested arrived every week. Nobody reopened it for five months. That is not negligence, it is the normal behaviour of an organisation: an exception once granted becomes a silent entitlement, and nothing in the process gives it an expiry date.

The reverse rule I set

Zero inventory submissions in the month scores zero on that criterion. No quality points granted on the strength of a count made in another month.

That rule is unpopular and it is necessary. Without it, a scoring system always ends up crediting the seniority of the last effort rather than this month's effort.

What a bonus system has to satisfy to exist

Someone has to reach it. Not everyone, but someone, and regularly. Over a period where only four winners emerge from twenty-seven available places, the question is no longer whether the teams are good, it is whether the grid is calibrated.

It has to exist in a single copy. One scale, one document, one version, one date. Three scales are worth zero scales, because each person can defend their own in good faith.

It has to aggregate the way it pays. If the bonus is monthly, the calculation is monthly. An average of capped averages does not say what a ratio over the period says.

And every exception has to carry a review date. Otherwise it becomes a permanent inequality nobody can justify any more.

Why this is a marketing matter

Because in a network, the scoring grid is the main internal brand document. It tells a hundred people what the company actually values, far more forcefully than a charter or an offsite.

A grid nobody reaches says one thing, and it says it every month: your effort does not count.