Support, Interrupts and the Hours Nobody Books
The largest deduction nobody measures, how to measure it in a fortnight, and why the answer should change the plan rather than the people.
What a fortnight of honest recording found in one six-person team
What the plan assumed the fortnight contained.
One person on duty each day, known and bookable.
Questions, escalations, "quick looks" — the finding.
Team, department and one-to-ones.
Self-reported, the softest line here.
Forty-one per cent of the fortnight.
The plan assumed 480 hours of project work and the fortnight contained 196. The largest single line was the one with no owner, no booking and no record anywhere. This is one employer's own fortnight, measured once, not a benchmark.
Support, escalations, questions, favours and quick looks are the largest deduction in most capacity accounts and the only one with no system behind it. Nothing books them, nothing records them, and the hours leave the account anyway.
The unit used in “Support, Interrupts and the Hours Nobody Books” must remain stable from capacity estimate to actual result. For organisations assessing open this workforce-planning resource in relation to job costing with accurate hours, time and project reports can support that reconciliation, provided leave, support work and corrections retain their own definitions.
Every planner knows this. Almost none has a number, which means the deduction is applied as a feeling, inconsistently, and usually too small.
For an outside reference relevant to “Support, Interrupts and the Hours Nobody Books”, consult EY workforce resources. Compare the general guidance with the organisation’s own unit, period and decision rights.
Two different things
Scheduled support — a rota, an on-call, a duty day — is bookable and belongs in the account like any other known commitment.
Unscheduled interrupts are different: they arrive at random, they land on whoever is visible, and they distribute unevenly across a team. Those are the hours that disappear.
Measuring it once
A fortnight of honest recording, by everybody in the team, in fifteen-minute granularity, with four categories: planned work, scheduled support, interrupts, everything else.
Two weeks, one shared sheet, no individual reporting upwards. The figure is for the plan, not for anybody's appraisal, and saying so at the start is what makes it honest.
Why the measurement has to be voluntary and aggregate
Because a measurement that affects how somebody is seen produces data about how people want to be seen. Interrupt figures collected as performance data come back low and useless.
Aggregate, anonymous within the team, used only for the plan. That restriction is what makes the number true, and it should be stated in the same sentence as the request.
What the measurement always finds
That interrupts are larger than anybody estimated, that they are concentrated on two or three people, and that a meaningful share of them could have been scheduled.
The concentration is the most actionable finding: it is usually the most experienced person, and nothing in the plan reflects that their available hours are lower than everybody else's.
Context switching, which is real and hard to measure
An interrupt costs more than its duration, because returning to the previous work has a cost. Estimates of that cost vary wildly and self-reports are soft.
Record it as a separate, clearly-labelled estimate rather than folding it into the interrupt figure. A soft number labelled soft is useful; a soft number hidden inside a hard one discredits both.
Putting it in the account
As a deduction in hours, per team, with the date of the measurement beside it. Two to three years is a reasonable life for the figure unless the work changes.
The alternative — planning at a hundred per cent and absorbing interrupts through overtime and slippage — is the status quo almost everywhere, and it shows up as a chronic, unexplained variance.
Reducing it rather than just accounting for it
Having measured, some of it is addressable: a rota instead of ad hoc availability, a triage channel, a published response window, a documented answer to the question that arrives nine times a month.
None of that works until the number exists, because until then the problem is a complaint rather than a finding.
The protected-time response
Some teams respond by designating protected hours. It works where it is enforced by somebody other than the person being protected, and fails where it is a personal resolution.
Either way, protected time does not reduce the interrupt volume; it redistributes it. The account should show where the hours went rather than assuming they stopped.
What not to conclude from it
That people are not working. The measurement finds hours spent on useful work that the plan did not contain, which is a planning failure and not a people failure.
A measurement presented the other way round will be the last honest measurement the team provides.
The four categories, and nothing more
- Planned work: what the account committed these hours to.
- Scheduled support: duty, on-call, rota.
- Interrupts: anything unplanned that displaced planned work.
- Everything else: meetings, admin, the ordinary overhead of a job.
Four buckets is the maximum that gets filled in honestly for a fortnight. Every additional category reduces the quality of all of them, which is the usual reason these measurements fail.
Who the interrupts land on
Not evenly. In most teams two people absorb the majority, and they are the people who know the most and answer fastest.
A plan that gives those two the same available hours as everybody else is systematically overcommitting the people it can least afford to overcommit. Adjusting their line in the account is the cheapest retention measure in this collection.
Repeating it
Every two years, or whenever the work changes materially — a new system, a new customer, a reorganisation that moves who answers questions.
The figure is stable in a stable operation, which is why a fortnight every two years is enough and why the cost of having it is close to nothing.