The Shrinkage Figure, Recalculated
When to recalculate the deductions in a capacity account, which lines go stale fastest, and how to do the recalculation without a measurement project.
One team's shrinkage, as planned and as it turned out over a year
Close; the entitlement is knowable.
Exact, as it should be.
Rate was two years old.
Two unplanned mandatory modules.
The measurement was from before the new service.
Six points, or about four days a quarter per person.
Two of the five lines were stale rather than wrong, and together they understated shrinkage by six points. Both had been measured once and never revisited. This is one employer's own year, not a benchmark.
Shrinkage figures go stale. The work changes, the team changes, a new service starts taking support, and the deductions that were measured three years ago describe an organisation that no longer exists.
The reconciliation in “The Shrinkage Figure, Recalculated” is strongest when assumptions and current work records can be examined separately. A team considering a structured route to remote desktop monitoring software for remote desktop monitoring software can add time and project context, with transparent settings, limited access and a clear correction route around consequential decisions.
Recalculating them is an annual exercise of a day or two, and skipping it is the most common reason a carefully-built account drifts into being wrong.
Additional context for “The Shrinkage Figure, Recalculated” is available from ASQ quality resources. Use external material to prompt a specific validation check rather than replace local judgement.
Which lines go stale fastest
Support and interrupts, by a wide margin: they follow the work, and the work changes continuously. Then sickness, which follows the workforce. Then training, which follows whatever the organisation has decided to be compliant about this year.
Leave and public holidays do not go stale at all; they are read rather than measured.
The annual recalculation
Leave: recompute from entitlements, which takes a query. Public holidays: read the calendar. Sickness: pull last year's rate. Training: ask the booking functions for their calendars.
Support and interrupts: re-measure, for a fortnight, the way it was measured the first time.
Why support needs re-measuring rather than re-estimating
Because it is the only line nobody can look up, and because it is the line that moves most. An estimate of a figure that moved by four points will be wrong by four points, and the estimate will feel confident.
A fortnight every two years is the cost. It is the single best-value measurement in capacity planning and it is the one most often skipped.
Recalculating does not mean rebuilding
Most lines will not move. The exercise is to check each one against its source and change the two or three that have, with a date.
A full rebuild each year is both unnecessary and risky: it tends to produce new definitions, which breaks the series and loses the comparability that makes the series useful.
When to recalculate out of cycle
A new service, a reorganisation, a significant change in headcount, or four periods of consistent variance on one line.
The last is the trigger that the variance series provides, and it is the reason to keep the series at all.
What a changed figure does downstream
Everything: net capacity, the difference, the committed position, and every plan built on them. A six-point shrinkage change is about four days a quarter per person, which in a team of six is most of a month.
That is why the change gets a date and a note, and why the plans published before and after are not directly comparable.
Telling people the figure moved
Plainly, with the evidence, in the account's header and in the first publication after the change.
An unannounced shrinkage change looks, from outside, like the planner adjusting the numbers to produce a desired answer — which is exactly what it would be if it were unannounced and unevidenced.
The figure that moves the other way
Occasionally shrinkage falls: support is automated, a service moves away, a training requirement is retired.
Applying that promptly matters as much as the increases, and it is the correction most likely to be delayed, because it makes the plan look more comfortable and nobody is in a hurry to appear less constrained.
Benchmarks, and why not to use them
Published shrinkage figures exist for several industries and they describe other organisations' work, other service models and other support arrangements.
Your own measured figure, however rough, beats somebody else's precise one. The only legitimate use of a benchmark is as a sanity check on an answer that looks implausible.
Who owns the recalculation
The account owner, with one person from each function that holds a source: payroll for leave and sickness, the training function for statutory requirements, and the service manager for support.
Four short conversations rather than a project. Nobody is being asked to produce a figure they do not already have; they are being asked to read it out.
Keeping the old figure visible
The superseded figure stays in the account with its date, which is what makes the series readable and what lets somebody reconstruct why last year's plan said what it said.
Overwriting it saves a line and costs the ability to answer the only question anybody asks about a changed number, which is what it used to be.
The two-figure year
Where a recalculation lands mid-year, the honest presentation is two figures with a boundary date rather than one blended figure.
A blended figure is a number that was never true at any point in the year and cannot be used to check anything. Two dated figures are both true and both checkable.
The annual line
"Shrinkage recalculated in March: 39.9%, up from 34.0%. Changes were sickness (+1.3) and support (+3.9); support re-measured over a fortnight in February."
Two sentences a year. They keep the account honest and the series readable, and they are the entire maintenance requirement for the supply side.