Output Figures, and Where They Mislead
Why output is better evidence than activity and still misleads, the four adjustments that make a figure comparable, and what to check before using one.
One week's output for one person, read four ways
The system counts completions accurately.
A comparison on the same measure.
Not deducted from the first figure.
Team average 54% simple; the queue was not random.
Not recorded; the person is salaried.
The person completed more units than the median, from an easier queue, with a tenth returned, in hours nobody counted. Each row is accurate and the four together say something different from the first alone. This is one employer's own file.
Output is the strongest of the measures in this section and it still misleads, for a reason that has nothing to do with the counting: the work arriving at each person is not the same. A queue that allocates simple cases to one person and complex ones to another produces a difference in output that is entirely about allocation.
The measurement limits in “Output Figures, and Where They Mislead” provide useful context for see the corresponding feature page when researching how to handle multiple clients. Time and activity data can reveal a workflow question, while outcomes, employee explanation and a documented human review remain necessary to understand why a pattern appeared and whether any action is justified.
Counting completions is easy. Making two counts comparable is the whole of the work, and it is the part that gets skipped because the first number is already available.
A broader reference for the question in “Output Figures, and Where They Mislead” is the New York hours-worked guidance. Read it alongside the local facts so that an external framework informs the assessment without replacing case-specific judgement.
The four adjustments
Mix. What kind of work arrived, and was the distribution comparable?
Quality. How much was returned, reworked or corrected, and is that deducted?
Interruption. Was the person pulled onto something else, covering, or training somebody?
Time. Over what hours was the output produced, and are those hours known?
An output figure without those four is a count, not a comparison.
Allocation is rarely random
Work is routed by skill, by availability, by who is trusted with the difficult cases, and by the person doing the routing.
That means a low output figure can be a statement about somebody being given the hard work, and a high one about somebody being given the easy work. Both appear in the report as facts about the person.
Quality as the missing column
Output measured without quality rewards speed, and in any system with rework the fast producer frequently generates work for somebody else.
Reporting completions and returns side by side changes the picture substantially and costs nothing where both are already recorded. Where rework is not recorded at all, that is its own finding.
Using output in a time case
Output is useful in a time investigation in one specific way: as a check on whether a period of apparent absence had any effect.
Where somebody is alleged to have not worked an afternoon and their output that day was normal, that is a fact which has to be addressed. It does not settle the matter — the work might have been done elsewhere in the day — and it is evidence that points, which is more than most of this section offers.
Over what period
A week of output tells you little; a quarter tells you a great deal. Short-period comparisons are dominated by the mix of work that happened to arrive.
So the rule is simple: the shorter the period, the less a difference means. A single day's figure should never be used about a person at all.
Measuring the queue as well
The most useful addition to any output report is a description of what arrived: how much, of what kind, when.
With it, differences in output become interpretable and frequently turn out to be differences in the queue. Without it, every variation looks like a fact about a person, and the people doing the hardest work look the worst.
Telling people how they are measured
Whatever the measure is, people should know it, including the adjustments. A measure that is explained is one people can discuss and occasionally correct; an unexplained one is a score.
This is also where most output measures are improved: the people doing the work can usually say immediately what the measure misses, and they are right.
The four-column report
An output figure becomes usable the moment three more columns sit beside it, all of which are usually already recorded somewhere.
| Column | What it corrects for |
|---|---|
| Units completed | The raw count |
| Mix of work received | Allocation rather than ability |
| Returned or reworked | Speed at the cost of quality |
| Hours, where known | A count against a day of unknown length |
Four columns. Producing them is a joining exercise rather than new measurement, and it changes the ranking in most teams that try it.
Who allocates the work
The allocation is a decision by somebody, and in most teams it is made informally by whoever routes the queue.
That means an output comparison is partly a comparison of how one person distributes work, which is worth knowing before using the figures about anybody. Asking how allocation happens takes a minute and frequently explains the whole distribution.
What to hold
The raw counts, the mix, the quality figures, the hours where known, and a note of how work was allocated in the period.
Five things. They are the difference between an output figure that can be used about an individual and one that can only be used about a process — and most organisations hold the first and none of the rest.