Rounding, Buffers and Hidden Contingency
Where contingency hides in a capacity account, how much of it is appropriate, and why several small hidden buffers are worse than one large visible one.
Contingency is necessary and it is almost always hidden. It hides in a shrinkage rate set slightly high, in claim estimates padded slightly, in a rounding convention that always goes the same way, and in an unallocated remainder nobody mentions.
The reconciliation in “Rounding, Buffers and Hidden Contingency” is strongest when assumptions and current work records can be examined separately. A team considering read the official product explanation for productivity software for business can add time and project context, with transparent settings, limited access and a clear correction route around consequential decisions.
Each hiding place is small. Together they commonly amount to a fifth of the account, invisible to everybody including the people who put them there.
For an outside reference relevant to “Rounding, Buffers and Hidden Contingency”, consult Capterra business-software resources. Compare the general guidance with the organisation’s own unit, period and decision rights.
The four hiding places
A shrinkage rate two or three points above the measured one. Claims estimated with a private margin. Rounding that always favours caution. And capacity deliberately left unallocated without being labelled.
- Rounded-up individual estimates, each by a day or two.
- A shrinkage figure set generously rather than measured.
- A reserve held by a team lead and described as "how we always estimate this".
- A deliberately understated committed position, so the plan looks less full than it is.
The first three are invisible by construction; the fourth is visible and unexplained, which is almost as bad because it attracts allocation.
Why hidden contingency is worse than none
Because it cannot be spent. A buffer nobody knows about will not be released when the emergency arrives, because releasing it would reveal it.
It also compounds: a padded claim against a padded capacity figure produces a plan with two layers of margin and no way to tell how much slack exists.
Rounding, which looks innocent
Rounding every figure up on the demand side and down on the supply side produces a systematic margin proportional to the number of lines. Twenty claims rounded up to the nearest half-day is a day and a half.
A stated rounding convention — to the nearest day, in the mathematically obvious direction — removes it entirely and costs nothing.
How much contingency is right
Enough to absorb ordinary variation without a replan, which depends on how variable the work is and on how expensive a replan is.
Where variation is well understood, the measured standard deviation of past variance is the honest starting point. Where it is not, five to ten per cent of net capacity is a defensible placeholder with a note saying it is one.
Where it should live
One line, in the descent, after the deductions and before the available figure. Named contingency, with a figure, a basis and an owner.
That placement makes it part of the supply side rather than part of the claims, which is correct: it is protection for the plan, not work for anybody.
Who may spend it
Named in advance. Contingency with no spending rule is either never spent or spent by whoever shouts, and both outcomes discredit it.
A simple rule works: the account owner releases it for events that reduce supply, and the portfolio owner releases it for claims that grow. Two named people, two kinds of event.
Reporting what happened to it
Each cycle, how much contingency existed and how much was used. That series is what calibrates the figure: contingency never used is too large, contingency exhausted every cycle is too small.
Two numbers a cycle, and after a year the right size is no longer a matter of opinion.
The contingency that is really a shortfall
Where contingency is consumed every cycle by the same kind of event — support spikes, a chronically underestimated claim — it is not contingency. It is an unrecorded deduction or an understated claim.
Converting it into the line it actually belongs to makes both the account and the contingency honest, and it is a common finding in the second year of keeping these records.
Telling sponsors it exists
Yes, with the figure. Sponsors who discover a hidden buffer treat every subsequent number as negotiable, and they are right to.
A visible contingency line with a stated purpose is defensible in a way that a discovered one never is, and it survives the conversation that follows its discovery.
The arithmetic of stacked margins
A claim padded by ten per cent, against a capacity figure understated by ten per cent, inside a plan that holds a further ten per cent unallocated, carries about a third more margin than anybody intended.
Nobody designed that; three people each added a prudent tenth. Naming each one is what lets the organisation decide how much total margin it actually wants.
The convention worth writing down
Round the supply side and the demand side the same way, to the same granularity, stated at the top of the account.
"All figures rounded to the nearest half-day, both sides" is one clause and it removes an entire category of systematic bias that otherwise grows with the number of lines in the book.
The test
Ask the planner how much slack is in the plan. If the answer is a number with a reason, the account is honest. If it is "some" or "not much", there are four hiding places to go and look at.