Practical guide
Document Interrupted Work Without Double-Counting Time
Trace where a task stopped, what work intervened and when it resumed. An original timeline and checked example separate effort, waiting, duplicate messages and unfinished output.
Track the task’s state, not every alert
Record the work that stopped, the request or condition that changed it, the work done instead, and the point at which the original task resumed or remained unfinished. A notification count cannot show that a task was interrupted, and elapsed time until completion is not automatically effort spent on that task.
HSE’s Demands standard concerns workload and work patterns. Its Control and Role standards concern influence over work and clear compatible requirements. Those references support asking how interruptions are assigned and handled. They do not provide a scientifically fixed restart penalty or validate this original logging method.
Use an existing approved task system where it can answer the question. A short record of one affected work window can be enough to begin a discussion. This is a work-planning aid, not surveillance of colleagues, an assessment of your speed or a burnout test.
Link each interruption to a work event
On a narrow screen, scroll the table sideways to read all columns. You can also focus it and use the arrow keys.
| Entry | What to record | Avoid |
|---|---|---|
| Original output | Current state, next step and delivery constraint. | “Busy morning” without the work affected. |
| Change event | Request identifier, arrival/awareness time and source category. | Counting three copies of the same request as three tasks. |
| Intervening work | Actual response, other assigned work or waiting condition. | Assuming every minute after a message belongs to it. |
| Resumption | Time or status, saved stopping point and next required step. | Marking a task completed because it was reopened. |
| Decision | Who may change routing, cover, sequence or deadline. | Changing required alert handling without authority. |
Arrival and awareness can differ. If you notice a message at 09:20 but it arrived earlier, retain both only when the distinction is useful. A record should say what you know rather than fill unknown times with false precision. Use categories such as routine request, required incident or clarification, according to the real procedure.
If a request arrives through chat and email, give it one work identifier with two notification references. Another reply on the same issue may create more work; record that actual work separately without inventing an additional deliverable.
Check a fictional non-overlapping timeline
The following times are an original fictional example, not measured productivity data or recommended work intervals. One worker has a 09:00–11:00 window. A report is still unfinished at 11:00.
On a narrow screen, scroll the table sideways to read all columns. You can also focus it and use the arrow keys.
| Window | Primary activity recorded | Minutes |
|---|---|---|
| 09:00–09:20 | Report drafting | 20 |
| 09:20–09:35 | Respond to one separate work request | 15 |
| 09:35–10:00 | Report drafting resumes | 25 |
| 10:00–10:30 | Already scheduled meeting | 30 |
| 10:30–10:40 | Input missing; recorded waiting, no other active work in this example | 10 |
| 10:40–11:00 | Report drafting resumes; output not complete | 20 |
The window contains 120 minutes: 20 + 15 + 25 + 30 + 10 + 20. Recorded report effort is 20 + 25 + 20 = 65 minutes. Other active work is 15 + 30 = 45 minutes; the remaining ten minutes are recorded waiting. Do not add the whole 120-minute window to the 65-minute effort total.
Suppose the separate request produced three messages. The example still contains one request and 15 minutes of recorded response work, not three 15-minute jobs. The scheduled meeting is another commitment, rather than automatically an interruption caused by the request.
Source: [1].
Keep overlapping and uncertain time honest
Real work does not always split neatly. A worker may monitor a queue while waiting for an export, or answer a message during a meeting. If you cannot allocate those minutes reliably, mark them mixed or uncertain. Do not credit the full same interval to two activities and present the sum as exclusive hours.
Check start and end times against the work window. Adjacent intervals can share a boundary without overlapping; an interval ending 09:35 and another beginning 09:35 do not create an extra minute. If your notes cover only part of a morning, do not extrapolate them into a full-week percentage without stating that missing information.
Restart effort may include finding the last version or reconstructing a decision. Describe that observed step and record its time only if known. This page gives no standard number of minutes to multiply by every interruption, and no estimate of the health or financial cost.
Source: [1].
Ask which work arrangement could change
Connect the record to a decision. Were routine requests duplicated across channels? Did a required incident need cover? Was the original task waiting for an authorized input? These can require different changes: one agreed request route, allocated queue cover, an input owner or a revised delivery plan.
An original conversation draft is: “The report received 65 minutes of active drafting in this recorded window. A separate request used 15 minutes, the scheduled meeting 30, and a missing input left ten recorded waiting minutes. The report remains unfinished. Could we review which requests interrupt report work, who covers the required route and what delivery estimate we can now confirm?”
This asks about the work arrangement instead of claiming that colleagues stole a whole morning. Use response-route examples to agree monitoring and genuine exceptions, and the capacity conversation for the revised workload calculation.
Review the output and route, with the limits intact
Keep the proposed change separate from its confirmation. After an appropriate work review, check whether the task had the agreed inputs, the required route remained covered and the expected output was delivered. Fewer notifications alone do not demonstrate any of those outcomes.
Use task labels and permitted references in a general record. Avoid collecting colleague response rankings, confidential message contents or health details simply to make the log look comprehensive. A concern about harmful behaviour needs its proper reporting route, rather than being buried in an interruption spreadsheet.
WHO describes burnout as an occupational phenomenon, not a medical condition. This timeline cannot diagnose it, assign severity or prove that a routing change improved health. Appropriate health, safety or workplace support need not wait for a completed diary or a particular score.
Use the original editable record
Download the original record (TXT). Fill your actual facts in your own copy; this page collects no answers.
INTERRUPTED WORK — ORIGINAL EVENT AND TIME RECORD Work window/date/time zone: Original task/current state/next step: Event/request identifier and source category: Arrival/awareness time; exact or approximate: Work stopped at; stopping point saved: Intervening work intervals and their sources: Duplicate notification references, not duplicate task counts: Waiting/mixed/uncertain intervals: Resumed at; completed or still unfinished: Non-overlapping active effort; elapsed window kept separate: Missing input/required route/coverage question: Person authorized to change the arrangement: Proposal/confirmation/review result: No fixed restart-time multiplier or diagnostic score.
Three checks before using your record
In the fictional two-hour window, was the report worked on for 120 minutes?
No. Active report intervals total 20 + 25 + 20 = 65 minutes. Other active work totals 15 + 30 = 45 minutes, and waiting totals 10 minutes. The elapsed window is 120 minutes; the report is still unfinished at 11:00.
One 15-minute request generates three notifications. Does that establish 45 minutes of work?
No. Link the duplicate notifications to the same event. The illustrated active interval remains 15 minutes; notification count does not multiply the duration.
I know two activities overlapped, but cannot reconstruct their shares. Should I charge the entire interval to each?
No. Mark it as mixed or uncertain, with the source and limits of the recollection. Do not count the full overlapping interval twice or invent exact allocation or a fixed restart penalty.
Sources and further reading
- HSE — Management Standards: Demands (11 May 2021)UK work-design standard: adequate and achievable demands within agreed hours; skills/capabilities and response to concerns. No meeting quota, clinical cutoff or prescribed tracking interval.
- HSE — Management Standards: Control (11 May 2021)A say in work patterns, pace where possible, and how work is done. Does not authorize unilateral changes to service or safety requirements.
- HSE — Management Standards: Role (11 May 2021)Clear, understood and compatible requirements and ways to raise uncertainty. No authority granted by this record.
- HSE — Management Standards: Support (11 May 2021)Usable support/resources and regular constructive feedback. Does not prescribe recognition categories or determine a pay entitlement.
- WHO — Burn-out as an occupational phenomenon (28 May 2019)ICD-11 occupational-context definition; not classified as a medical condition. No diagnosis, severity score or treatment selection.