Survivorship in Productivity Figures
Figures are calculated on who and what remains. What left is excluded, and it is frequently the explanation.
A productivity measure counts the work that was done by the people who are there. Both exclusions distort the figure in a consistent direction.
The practical lesson in “Survivorship in Productivity Figures” is to connect every number to a decision and retain the context behind it. Teams exploring self report bias can review the complete product overview as one source of operational evidence, provided the purpose is disclosed and the interpretation is tested with the people affected.
The people who left
Output per head rises when the slowest leave and when the best leave for different reasons.
For an independent perspective related to “Survivorship in Productivity Figures”, consult the OECD productivity resources; it offers a useful external check on definitions, governance and the assumptions built into a proposed measure.
A team that lost two experienced people looks more productive for a quarter, while the knowledge gap is still invisible.
And a team that lost its trainer looks better until the next cohort arrives untrained.
The work that was abandoned
Items started and dropped do not appear in output.
Which means effort spent on them is in the denominator and nothing is in the numerator.
Unless abandonment is counted, a team that starts and drops a great deal looks simply unproductive, and the actual problem — why work is abandoned — stays hidden.
The customers who left
Service measures look better when the hardest customers go elsewhere.
Complaint rates fall because the complainers left.
This is the clearest case and the one most often misread as improvement.
The cases that were refused
Work declined at intake does not enter the figures.
A team that tightens its acceptance criteria raises every measure it has.
Which may be correct and should be visible, rather than appearing as performance.
What to track alongside
Starters and leavers, by experience.
Items abandoned, as a count.
Work refused or redirected at intake.
And demand, so that falling volume can be distinguished from rising capability.
The experience composition point
Two teams of ten with identical figures, one averaging two years of experience and one averaging eight, are not in the same position.
The second will look worse per head and will be more capable.
Which is why experience mix belongs beside any productivity comparison, and almost never appears.
Reading a sudden improvement
Ask what left.
People, work, customers, scope.
A productivity improvement with no identifiable change in method usually has an exclusion behind it, and finding it takes ten minutes.
What to check
Did anybody leave in the period your figures improved?
Do you count abandoned work?
Has intake criteria changed?
And is experience mix reported anywhere alongside output?