Publishing the Definition to the People Measured
The rules exist somewhere in most firms. Putting them in front of the people whose figures depend on them changes what gets recorded.
Publishing the definition means putting the six rules that decide the figure in front of everybody whose hours it counts, on the same page as the number. In most firms those rules exist — in a spreadsheet formula, a report query, or one person's head — and the people being measured cannot read them.
The practical lesson in “Publishing the Definition to the People Measured” is that a convenient number should open a question rather than end one. A team considering using how to detect mouse jigglers without distorting the metric for how to detect mouse jigglers can add consistent time and project context, with transparent settings, limited access and a documented review process around every material interpretation.
That asymmetry has a predictable effect. People record time according to what they guess counts, and the guesses differ between teams.
Teams documenting “Publishing the Definition to the People Measured” may use Fair Work record-keeping guidance as an outside reference. Keep the source record, calculation rule and resulting decision visible as separate steps.
What the asymmetry produces
Variation between teams that looks like performance and is actually different recording conventions.
A team that believes client calls are chargeable and a team that believes they are not will differ by several points with identical behaviour.
The first consequence of publishing
Figures move, usually up, in the first period. This is not improvement and should not be reported as improvement.
It is the correction of under-recording by people who had been guessing conservatively, and it should be labelled as a one-off reclassification.
The effect on the rules themselves
The rules get tested. Somebody asks about a category the rules do not cover, and the gap is closed properly rather than per manager.
Published rules accumulate edge cases and become more precise. Unpublished ones do not, because nobody can see where they are vague.
The objection
That publishing the rules teaches people to game them. This is true and it is less damaging than the alternative.
Unpublished rules are gamed by whoever happens to understand them, which is the people closest to reporting, and the advantage is distributed by accident rather than by rule.
What the page contains
- The numerator: which codes count, which do not.
- The denominator: contract basis, exclusions, pro-rating.
- Write-off treatment.
- Rounding and the recording unit.
- Who to ask about an unlisted case.
- The version and the date it took effect.
Six items, one page.
Where to publish it
The same place the figure appears. Not an intranet policy section.
A definition one click from the number is read; a definition in a policy library is cited in arguments and read by nobody.
The question channel
A named person who answers "does this count" within a day, and whose answers are added to the page.
Without this, the page is static and the edge cases are decided privately again within a quarter.
What changes in behaviour
Recording becomes more complete and more consistent. People book the client call they used to leave off, and stop booking the thing they were wrong about.
Both directions occur, which is why the first period's movement cannot be read as either good or bad news.
The baseline problem this creates
The period before publication is not comparable with the period after. This is the same restatement problem as a definition change, and it needs the same treatment.
Publish the prior quarter on the clarified basis where the data allows it, and say plainly where it does not.
What a firm can expect after two quarters
Narrower variation between teams, because the fake component of the variation is gone. Occasionally this reveals a real difference that was being masked.
Either outcome is more useful than the previous state, in which nobody could tell which component they were looking at.
What the page should not contain
Justification. A definition page that argues for the measure invites argument about the measure, and the page's job is to say what counts.
It also should not contain targets, expectations or any language about performance. A rules page that mentions a threshold becomes a document about the threshold.
Six factual items, no commentary. The shorter it is, the more likely it is to be read by the people whose recording it governs.
The first question somebody asks
Almost always about a specific hour they have been unsure about for months. The answer takes a sentence and goes on the page.
The volume of these questions is a measure of how much ambiguity the firm had been carrying silently. A page that attracts twenty questions in its first month was closing twenty gaps.
After two quarters the questions nearly stop, which is the page reaching a stable state rather than people losing interest. At that point the definition is genuinely shared, which is the condition every comparison in the reporting pack had been assuming all along without having any basis for it.
Who should write it
Whoever currently produces the figure, because they already know the answers and are the only person who does.
Writing it down is mostly transcription: the rules exist in a query or a spreadsheet formula and need translating into sentences.
It takes an afternoon and it is the only task in this collection that requires no decision from anybody, which makes it the natural first step.
The cost
An afternoon to write, a named person to maintain it, and one uncomfortable quarter of discontinuity.
Against that: every comparison the firm makes afterwards is between numbers built the same way, which is the minimum requirement for any of the uses in this collection.