The Quarter the Definition Changed
A firm moved internal meeting time out of the denominator and reported an eight-point improvement. The work did not change.
One quarter, reported and restated, after internal time left the denominator
1,418 available hours. The figure in the original pack.
1,290 available hours. The same work, measured the new way.
1,284 available hours. What the board was shown beside 79.2%.
1,412 available hours. What the scheduling change actually produced.
The reported movement was 9.2 points; on a consistent basis it was 1.3, and on the old basis 1.1. The difference is the denominator, and the firm presented it as the result of a scheduling programme. One firm's own quarter, not a benchmark.
A professional services firm reported utilisation of 79.2 per cent for one quarter and 88.4 for the next. The improvement was presented as the result of a scheduling initiative.
The practical lesson in “The Quarter the Definition Changed” is that a convenient number should open a question rather than end one. A team considering see the complete workflow summary for online timesheets for client billing can add consistent time and project context, with transparent settings, limited access and a documented review process around every material interpretation.
What had changed was the denominator. Formally committed internal time had been moved out of available hours between the two reports.
Additional context for “The Quarter the Definition Changed” is available from Canada Revenue Agency payroll guidance. Current legal, employment, tax and accounting requirements should be confirmed with a qualified adviser where necessary.
What was actually done
A reasonable thing. Staff had a fixed half-day a week of internal commitments, and counting that as available for client work made the figures unreachable.
Removing it was a correct decision, defensible on exactly the grounds set out in the denominator note here.
What went wrong
The prior period was not restated. The report showed 79.2 and 88.4 side by side as a trend.
Restated on the new basis, the first quarter was 87.1. The real movement was 1.3 points, not 9.2.
How the misreading spread
The nine-point figure went into a board pack and then into a staff communication describing the scheduling change as a success.
The scheduling change had in fact produced about a point, which is a reasonable result for one quarter and not the one that was claimed.
The consequence six months later
The initiative was extended on the strength of the reported result, and the second extension produced nothing measurable because the first had not produced what it was credited with.
Resources went to a programme whose effect size had been overstated by a factor of seven.
The second consequence
Targets were raised. The new baseline looked like eighty-eight, so ninety was set as the next step.
Under the old definition that was eighty-two, which nobody had claimed was reachable, and the staff who had to reach it could not explain why the goalposts had moved.
Why nobody caught it
The two changes were made by different people in different months. The denominator change was a data decision and the reporting was a management process.
Neither party saw both halves, which is the ordinary way this error occurs. It does not require anybody to be concealing anything.
The restatement rule
Any change to numerator or denominator requires the previous four periods restated on the new basis, published together with the old basis, in the same table.
Two columns for one quarter. It makes the size of the definitional change explicit and separates it from the operational change.
The version label
A definition that changes gets a version number that appears beside every figure produced under it.
A reader seeing "def. 2" against "def. 3" knows to ask for the restatement. A reader seeing two bare percentages does not.
What the firm did afterwards
Published both bases for a year, and found that the second year's genuine improvement was about four points, which was a real and reportable result.
The cost of the correction was one awkward meeting. The cost of not making it had been a misallocated programme and a target nobody could hit.
The test to run now
Pull the definition in force for each of the last eight quarters. If that information does not exist, the trend line in the current pack is of unknown reliability.
Most firms discover at least one undocumented change when they look.
How the two halves got separated
The denominator change was made by an analyst implementing a reasonable request. The reporting was prepared by a practice manager using the system's output.
Neither had any reason to think the other's work had changed. The analyst had fixed a fairness problem; the manager had produced the usual slide from the usual query.
That separation is the ordinary condition in a firm of any size, which is why restatement has to be a property of the reporting process rather than a courtesy somebody remembers.
What a restatement looks like on the page
Two columns for the current period and the four before it: old basis and new basis. A line naming what changed.
Six numbers more than the usual table and one sentence. It fits in the same space.
Readers handle it without difficulty, which is worth saying because the usual objection to restatement is that it will confuse the board. Boards deal with restated figures routinely in financial reporting and find them unremarkable; the confusion is anticipated rather than observed, and the alternative is a trend line that silently includes a nine-point definitional step.
The one-sentence rule
No change to the numerator or denominator ships without the prior four periods restated on the new basis in the same table.
One sentence, enforced by whoever produces the pack rather than by policy, and it prevents the entire failure described here.
The firm in this case adopted it afterwards and has used it twice since, both times finding that the definitional component was larger than the operational one.
The general shape
A measure can improve because the world changed or because the measurement changed, and the two are indistinguishable from the reported number alone.
Only restatement separates them, which is why it is the one non-negotiable rule in this section.