Time Observation Sheet Template for Service Operations

A time observation sheet records the steps and intervals you actually see while work moves through a process. For a service team, that means documenting active handling, waiting, rework and missing evidence—not simply writing down how long the whole request took.

The copy-ready resource below includes a study plan, interval ledger, case summary and review record. It is designed for a small process-improvement observation, not for setting pay, rating individual workers or establishing an engineered standard time. Everything needed to start is inline; no download or specialist software is required.

Define what the sheet will measure

Choose one case type and one question. For example: “Where does a routine service-address correction wait between acceptance and confirmation?” Do not combine routine corrections with disputed requests or other work that follows materially different steps.

This service version focuses on observed process behavior and supplies its own interval categories and review rules. It does not reproduce a manufacturing performance-rating worksheet.

Use a case clock, not an employee clock

Follow one request from the agreed start event to the agreed finish event. Keep the elapsed clock running during waits. Record a new row whenever the request changes activity or state. Record elapsed minutes for the case, not the sum of everyone’s labor.

Use these mutually exclusive categories for the primary case timeline. They are local operating definitions for this resource, not universal Lean classifications:

• Handle: First-pass work that advances the case, including a required review.
• Wait: No one is actively advancing this case; it awaits information, a person, a system or a scheduled event.
• Rework: Active repetition or correction of an earlier case step because its previous output was inadequate.
• Unknown: You cannot establish what happened during the interval.

Do not call every review “waste,” or equate Handle with customer value. A required check may be necessary even if the customer would prefer not to wait for it. Establish the facts before deciding what can change.

Copy the observation plan and ledger

Copy the following blocks into a shared document or spreadsheet. Use one plan per study version, one summary per case and as many ledger rows as the observation requires. Replace bracketed placeholders; do not delete an inconvenient row to make the totals look cleaner.

Study plan

STUDY ID / VERSION: [unique reference; version; effective date]
QUESTION: [one process question the observations should inform]
CASE TYPE / EXCLUSIONS: [eligible work; excluded routes and why]
PROCESS OWNER: [named decision-maker; backup]
OBSERVER / REVIEWER: [named recorder; independent checker]
PERMISSION / BRIEFING: [owner authorization; staff informed; date]
START EVENT: [observable event beginning the case clock]
FINISH EVENT: [observable event ending the case clock]
TIME BASIS: [calendar elapsed minutes; timezone; timestamp precision]
OBSERVATION WINDOW: [dates, sessions, locations or teams]
CASE SELECTION: [how cases enter; record every selected case]
CAPTURE METHOD: [direct observation; permitted event records]
CATEGORY RULES: [Handle / Wait / Rework / Unknown definitions]
PARALLEL RULE: [split/flag mixed activity; no double-counted elapsed time]
EVIDENCE RULE: [restricted reference IDs; no unnecessary personal data]
REVIEW DEADLINE: [named checker; explicit date/time or due rule]
ESCALATION: [owner/backup; when and how unresolved items are raised]
RELEASE RULE: [checks required; allowed conclusions; limitations]

Case interval ledger

study_id,case_id,row_id,step,state,started_at,ended_at,elapsed_minutes,role,evidence_ref,observation_note,issue_flag
[study],[case],[unique row],[observable step],[Handle|Wait|Rework|Unknown],[timestamp with offset],[timestamp with offset],[end minus start in minutes],[role not personal rating],[reference],[fact not assumed cause],[none|gap|overlap|mixed|estimated|other]

Each row needs a case ID, unique row ID, recognizable step, category, start, end, duration and capture basis. An evidence reference can point to the observer’s dated notes; it need not be a screenshot. Use restricted case aliases rather than copying customer messages or personal details into a broadly shared worksheet.

Use timestamps with dates and UTC offsets, such as 2026-11-12T09:00:00-06:00. If timestamps have different offsets, calculate their difference as instants, not by subtracting the displayed clock digits. Use one precision consistently; this example records whole minutes.

Case summary and decision record

CASE ID / STUDY VERSION: [references]
OBSERVATION STATUS: [Complete | Partial | Cancelled]
START / FINISH OR CUTOFF: [timestamps; end event or observation cutoff]
HANDLE / WAIT / REWORK / UNKNOWN MINUTES: [separate sums]
TOTAL OBSERVED MINUTES: [sum of all four categories]
BOUNDARY MINUTES: [finish or cutoff minus start]
RECONCILIATION DIFFERENCE: [total observed minus boundary]
WAIT SHARE: [100 × Wait / total observed; N/A for zero duration]
ISSUES / LIMITATIONS: [gaps, ambiguity, route differences, capture source]
REVIEW STATUS: [Draft | In review | Blocked | Accepted]
REVIEWER / TIME / EVIDENCE: [checker; timestamp; reviewed record]
OWNER DECISION: [further observation or specific improvement experiment]
ACTION OWNER / DUE / FOLLOW-UP: [person; exact deadline; review date]

Capture and validate the intervals

1. Brief the people involved. The process owner approves the scope and tells staff what will be observed and why. Explain that the purpose is to understand the workflow, not rank people. Agree on permitted records and access before collecting anything.
2. Pilot the event boundaries. Observer and reviewer classify one practice case together. Resolve what counts as acceptance, completion, rework and waiting. If they disagree, revise the study plan before collecting the study cases.
3. Record in sequence. Start a row at the start event, close it at the next state change and open the next row at that same instant. The Gemba Academy facilitator guide likewise emphasizes understanding work elements and capturing timing cycle by cycle. Here, follow each service case chronologically rather than reconstructing a preferred sequence afterward.
4. Validate before review. End must not precede start; durations must match timestamps; row IDs must be unique. Within each case, adjacent rows should meet without unexplained gaps or overlaps. Retain observed deviations from the planned process.
5. Reconcile and classify. Add Handle, Wait, Rework and Unknown minutes. Their total must equal the case boundary duration. A difference of zero is necessary, but does not prove that every category is correct.

For this template, a case is eligible for a complete timing comparison only if its status is Complete, its reconciliation difference is zero, and all unknown, mixed or estimated intervals have been resolved with evidence. Keep other selected cases in the study register with their limitations rather than silently dropping them.

Worked example: a service-address correction

Fictional example: An operations team observes one routine service-address correction on November 12, 2026. The clock starts when complete correction details are accepted and ends when confirmation is sent. Observer Noor records the case; reviewer Alex checks it; process owner Casey decides what to investigate. All displayed times below are America/Chicago, UTC−06:00, on that same date.

Handle = 4 + 6 + 3 = 13 minutes. Wait = 8 + 5 = 13 minutes. Rework = 3 + 4 = 7 minutes. Unknown = 0 minutes. The sum is 33 minutes, equal to 09:33 minus 09:00. Reconciliation difference = 33 − 33 = 0. Wait share = 13 ÷ 33 × 100 = 39.4%, rounded to one decimal.

The second reviewer check is Rework because the plan classifies repeat checking caused by an earlier inadequate output that way. The original required review remains Handle. This avoids changing the definition to make a preferred conclusion look true.

Alex reviews the seven notes and timestamp continuity by November 13 at 12:00 local time. If they agree, status becomes Accepted. Casey records a proposed test of entry validation by November 13 at 17:00, with Noor responsible for observing comparable cases after the change. These deadlines are illustrative team choices, not external standards.

One case does not show a typical wait share or establish the cause of the delay. The notes show two waits and a correction loop. Casey still needs to check whether those patterns recur and whether a proposed change preserves correction quality.

Handle gaps, parallel work and interrupted observations

Missing or uncertain time

If the observer leaves from 09:18 to 09:26 without a trustworthy event record, enter an eight-minute Unknown interval. Do not label all eight minutes Wait just because the case had not finished. The elapsed total can reconcile while the classification remains unresolved; mark review Blocked for complete timing comparison.

If authorized records later resolve the gap, retain the original observation, corrected version, evidence reference and reviewer decision. Label estimates as estimates; do not invent seconds to suggest greater precision.

Concurrent activity

If two people work simultaneously from 09:12 to 09:18, the case has six elapsed minutes, not twelve. A separate effort ledger may record twelve person-minutes, but do not add it to the case timeline. If concurrent activities belong to different categories, split them when evidence permits or flag a Mixed interval and block category comparison until the reviewer applies a documented rule.

Partial, cancelled and reopened cases

• Still open at the cutoff: Mark Partial, record the cutoff and reconcile the observed span only. Do not present that span as completed case duration.
• Cancelled: Record cancellation time and reason, retain the selected case and report it separately. Do not replace cancellation with the normal finish event.
• Reopened after confirmation: Keep the original complete observation. Start a linked follow-up episode if needed, and disclose the reopen when interpreting quality. Do not silently overwrite the original finish.
• Different route or batch: Record the route or batch ID. Separate unlike cases before summarizing; do not divide batch time among cases without an explicit allocation rule.

Review the evidence before changing the process

Use Draft while collecting, In review when the summary and references are ready, Blocked when a required field or timing issue is unresolved, and Accepted only after the checker signs off. The process owner handles disagreements; the observer should not approve their own unexplained adjustment.

A practical starting rule is reviewer check by the next business day after capture, using the team’s documented working calendar. If still blocked at that deadline, notify the process owner and name a recovery owner and next check time; the backup takes over if the owner is unavailable. Acceptance means the record passes its stated checks, not that the sample represents all future work.

Report selected, complete, partial, cancelled and blocked case counts alongside any summary. Compare like routes and observation conditions. There is no universal number of cases that makes this worksheet statistically representative. Decide the sampling and analysis method appropriate to the decision before using observations to set targets or staffing assumptions.

Once the evidence is validated, use it to inform a value stream map for the wider workflow. Keep the two tasks separate: the sheet captures what happened to observed cases; the map brings steps, information flows and improvement choices together.

Where Workhint fits in observation review

The sheet works in a shared document. Workhint is relevant when collecting the record, returning incomplete submissions and obtaining a review decision become recurring coordination work. The Workflow Engine describes defined request steps, role-based approval routing and required information before work proceeds.

The workflow automation solution describes forms and review stages with assigned approvers. Those capabilities offer a basis for configuring the human submission-and-review process around this resource. They do not establish that Workhint automatically times work, reconciles these intervals, performs statistical analysis or judges whether the observations are accurate. Confirm the specific configuration separately.

FAQ

Is this a time study or a timesheet?

It is a process observation record. It follows a case and classifies elapsed intervals. A timesheet records a person’s work time; a formal time study may also require performance rating and allowances that this resource deliberately does not supply.

Can system timestamps replace direct observation?

They can support the record if the event meanings and capture basis are verified. A ticket marked “in progress” does not prove continuous active handling. Keep uncertain intervals Unknown until evidence supports a category.

Should waiting stop the clock?

Not for this calendar-elapsed case timeline. Record waiting as its own interval. If the team also wants staffed-hours time, calculate and label it separately using a defined working calendar; do not quietly subtract nights or breaks.

Can I copy this into a spreadsheet?

Yes. Use the ledger headings as columns and keep the plan and summary alongside them. Calculate duration as the timestamp difference in minutes, validate the categories and preserve the review record. The inline resource is the complete template; no downloadable workbook is promised.