Time and Motion Study: A Two-Week Baseline Worksheet

TL;DR: Track one manual workflow for 10 business days. Record active work, waiting, rework, and unfinished cases separately. Use the baseline to test a buying case; it cannot prove future automation savings on its own.

The team says a task takes five minutes. The vendor says software can remove most of it. Neither number tells you how often people chase missing information, repeat a check, or leave a case open overnight.

Before you buy, collect a small record of what actually happens. This worksheet helps an owner and an operator agree on the current workload, the gaps in the evidence, and what a pilot must demonstrate. To start collecting, copy the four tab layouts below and use the six-step schedule.

What is a time and motion study before automation?

A time and motion study records the work people do and the time it takes. For this baseline, “motion” means the steps between receiving a request and accepting the finished result, including copying data and switching systems. It is a practical observation exercise, not a method for setting employee speed quotas.

A baseline is your starting measurement before the process changes. Workflow automation uses software to move work through defined steps; the baseline tells you which steps might be worth changing. The broader business process automation ROI guide explains how those measurements later become a business case.

Use one repeatable workflow with an observable finish. Here are five suitable boundaries:

This approach matters beyond software. In EPA Region 10's spring 2014 pilot, average correspondence closure time fell from 26 to 14 business days. The same pilot improved deadline performance from 28% to 58%. Those public results followed process changes, including triage and response templates; they are not a savings benchmark for your business.

What belongs in your baseline worksheet?

Measure both active human work and elapsed time, plus outcomes and rework. Active work shows labor demand; elapsed time shows how long a customer waits. The EPA metrics guide makes the same distinction between processing time and lead time.

Create a workbook called “Two-Week Baseline Worksheet: Measure the Manual Process Before You Automate It.” Use four tabs: Scope, Cases, Work Sessions, and Summary. The field lists below are the worksheet; copy each field into a column, then add observations as rows.

Scope tab: freeze the measurement rules

Fill this in before the first counted case. A customer relationship management system, or CRM, stores customer and sales records; its case ID can connect your worksheet to the original record. Keep customer names and message contents in the approved source system, not in the observation sheet.

Cases tab: one row per distinct item

Log every eligible arrival if practical. A “cohort” is the group of cases that entered during the same measurement window. Keep that group identifiable even when some cases finish later.

Work Sessions tab: one row per stretch of work

Record each person's active minutes against the Case ID. Classify a session as initial work, required review, or rework, so review is not hidden inside “saved” time later. Stop the active timer for unrelated interruptions; leave elapsed time running.

If two people work together for five minutes, record ten person-minutes. That is labor effort, not ten minutes of elapsed time. For batch work, record the batch time once and allocate it across the affected cases using a stated rule; do not charge the full batch to each row.

Do not calculate waiting by subtracting summed person-minutes from elapsed time when people work in parallel. Use timestamps for elapsed time and separate delay observations for waiting. Label an unknown delay as unknown.

Run the two-week baseline without changing the process

Plan the improvement by first defining the decision, collecting an unchanged baseline, and reviewing the evidence. Start with ten business days as a manageable collection window. Keep actual process improvements for the next phase so the before measurement remains interpretable.

The ASQ check-sheet procedure calls for clear definitions and a short trial before collection. Use that trial to discover confusing fields, then freeze the form. The following schedule is a suggested operating plan, not a statistical standard.

1. Before day one: agree on the buying question. Choose one workflow and its acceptance rule. Export the opening backlog from the source system. Record its size and case IDs separately from new arrivals.
2. Before day one: trial the sheet together. Have two people classify the same few examples. Resolve differences over “done,” “rework,” and timer pauses before counted collection begins.
3. Days one through four: capture normal work. Use your existing spreadsheet, a timer, and read-only exports from the CRM or ticket system. Join observations with Case ID; avoid installing a new integration just to collect the baseline.
4. Day five: check coverage. Match worksheet arrivals to the source system. Investigate missing IDs and impossible timestamps. Add corrections with a note; never fill missing work time with zero.
5. Days six through ten: keep the same rules. Include normal late shifts and difficult cases. Log outages, holidays, absences, and any necessary process changes. If a change alters the workflow, separate the periods or restart the affected measurement.
6. At close: reconcile and sign off. Save a fixed snapshot, list unfinished cases, and agree whether collection must continue. The operator checks the workflow facts; the owner checks what decisions the data can support.

Tell staff that the purpose is to understand work, not rank individuals. Otherwise, they may postpone difficult cases or skip breaks to make the exercise look good. Measure the logging effort separately so the worksheet does not make the manual process appear more expensive than it normally is.

Is your time study sample size good enough?

There is no universal number of observations that makes a baseline reliable. You need coverage of normal case types and enough data for the decision you face. Keep cases still open on day ten in the arrival cohort, report their age and unfinished status, and follow them under the agreed rule instead of assigning a zero completion time.

NIST's sample-size guidance ties the required count to variation and the precision needed, not a fixed number of days.

For a busy workflow, capture all arrivals and use a preselected sample for detailed timing if timing every case is too costly. Spread that sample across days, staff roles, and case groups. Record the selection rule and how many cases were eligible, selected, and actually timed; NIST's sampling-plan guidance emphasizes representative observations.

Compare the first and second weeks by case group. A larger share of complex work can raise the overall average even when each group works at the same speed. Do not multiply an oversampled exception group's time by the volume of all cases; weight group averages using the actual arrival mix.

Reconcile completed and unfinished cases

Keep work still open on day ten in the arrival cohort, with its current status and age. Its completion time is unknown, not zero. Report the completed subset separately, and continue following entered cases under the rule agreed in Scope before calling its average a full-cohort result.

Use this count check: opening backlog + eligible arrivals − accepted outcomes − other exits = closing backlog. Count accepted outcomes from both opening backlog and new arrivals in this check. For arrival-cohort timing, use only the new-arrival cohort; old backlog completions belong in a separate analysis.

Only use complete histories in the mean-effort row, including initial work, review, and rework. Keep effort on unfinished or unsuccessful cases visible in a separate total; it is still a cost. If you selected a timing sample, do not imply its effort total covers every arrival.

Extend the baseline when missing observations cluster on busy days, a major case type never appears, or month-end work falls outside the window. Zero observed errors does not establish a zero error rate. A weak sample is a reason to collect more evidence, not a reason to make the forecast look cleaner.

A time study calculation example

This operator composite shows how the worksheet changes a buying decision. It is an invented planning example, not a public customer claim or a measured That'sGonnaHelp result. All volumes, rates, and proposed pilot outcomes below are assumptions.

A small B2B service team wants to automate intake checking. Its manager estimates five minutes per request and receives about 240 requests per ten business days. That suggests 20 hours of manual work, but the estimate comes from memory and excludes return visits to incomplete records.

The team uses an existing CRM, a shared spreadsheet, and a timer. It freezes the start at receipt and the finish at an accepted intake record. Every eligible arrival gets a Cases row, while Work Sessions records initial handling, review, and corrections against the same ID.

At day ten, 216 new requests are accepted and 24 remain open. One operator initially leaves out email corrections; the daily check finds the gap, and source records support a correction to the log. Any minutes that cannot be recovered stay marked missing rather than becoming invented measurements.

In this hypothetical completed dataset, all 216 accepted cases have full timing records totaling 2,592 person-minutes, or 12 minutes each. Their median elapsed time is six calendar hours. The 24 open cases have consumed another 240 minutes so far; the team reports those four hours separately and waits for the cases to finish before estimating effort for the full arrival cohort.

Suppose follow-up later produces a complete cohort averaging 12 minutes across all 240 requests, with no remaining timing gaps. That is 48 hours per observed cohort. The proposed automation target is six minutes of human work per case, including review and exceptions; achieving it would release 24 hours at that volume, but the baseline has not demonstrated the target.

At an assumed $35 per hour, those 24 hours represent $840 of capacity value per comparable period. Suppose a pilot and setup cost $3,000 and ongoing costs are $500 per such period: the capacity-based net is $340, with about 8.8 periods of simple payback. If the hours do not reduce an expense, cash payback remains unproven; a later pilot must establish the after result and the owner must identify an actual use for the capacity.

Can you prove automation ROI before buying?

Before buying, you can substantiate the current workload and test an ROI forecast; you cannot prove savings that have not happened. Measure automation ROI later by comparing the same defined outcomes, case mix, and human effort before and after a pilot, then subtract the full project costs. Treat vendor claims about the after state as hypotheses. When deciding how to prove automation ROI before buying, separate the baseline you can verify from the future result you still need to test.

Your handoff should contain the Scope tab, the observation log, the reconciled Summary, and an assumption sheet. For each proposed saving, name the manual step it affects and the future measurement that would confirm it. Use the time-savings and loaded-labor guide when converting supported minutes into dollars.

Time study tools and USD cost

You need a spreadsheet, a timer, source records, and someone to review the evidence. LibreOffice Calc is free, so a new software license can cost $0; an already licensed business spreadsheet may also have no incremental license cost. The main baseline cost is staff time: the example below totals $297.50 for an assumed 8.5-hour study, before extended follow-up.

The amounts below are a worked planning budget in USD, not vendor quotes or market averages. Use your own labor rate and actual effort. Logging time belongs in this study budget, not in the normal workflow's per-case effort.

The rate range is an illustration, not a market benchmark. Multiple Work Sessions entries per case can exceed the assumed logging effort, and later follow-up adds time. Replace both the rate and entry count with your actual study inputs.

Private-industry employer compensation averaged $45.65 per hour worked in June 2025, including $13.58 in benefits. That BLS release, published September 12, 2025, is context for including benefits; it does not supply the right rate for your team. The example's $35 rate is a separate assumption.

Automation ROI metrics for the vendor handoff

Give each measured input a source, period, and confidence note. Keep expected future volume and proposed minutes saved in separate assumption columns. These are the minimum decision lines:

• Work removed: which recorded steps disappear, and which reviews remain?
• Quality retained: what acceptance check must the pilot still pass?
• Full operating cost: licenses, implementation, training, maintenance, review, and recovery.
• Benefit realization: what expense will fall, or what funded work will use the freed hours?
• Stop rule: what result would make you pause, redesign, or decline the purchase?

Enter the measured baseline and clearly labeled assumptions into an automation ROI calculator, then test a slower, more expensive case. Do not add “rework savings” again if rework is already part of the time reduction. For a marketing-specific forecast, the ROI assumptions checklist helps separate projected lift from observed operating data.

When this baseline is not a good fit

A two-week window is a poor fit for rare work, strongly seasonal work, or a process being redesigned during collection. Extend the window for rare events, measure the relevant season, or freeze a new process version before starting. This worksheet supports routine operating decisions; it does not certify safety or replace specialist evaluation of high-stakes decisions.

If the problem is unclear ownership or contradictory approval rules, fix the definition before measuring potential automation. A back-office workflow audit helps locate those handoff problems. Faster software cannot resolve an unanswered business decision.

Common mistakes

• Timing only the easy cases. Follow the selection plan and report missed cases.
• Counting a queue delay as paid labor. Keep elapsed time and human effort separate.
• Dropping unfinished work. Show open-case count, age, and effort spent so far.
• Changing the process halfway through. Split the observations by process version.
• Calling released hours cash savings. Name the expense change before claiming cash ROI.

FAQ

A baseline tells you what the manual process currently requires and where the evidence is incomplete. Use the answers below to handle buying and measurement questions the worksheet alone cannot settle.

Can two weeks prove the ROI of automation?

No. Two weeks may establish a useful starting point for frequent, stable work, but the automated result remains unmeasured. Use the baseline to set pilot targets, then compare outcomes after the pilot under the same rules.

How do you calculate automation ROI?

For a chosen period, calculate (realized benefits − total costs) ÷ total costs × 100. Include implementation and operating costs for that period, state how benefits become value, and use a matching time horizon. Report a capacity-based estimate separately when payroll or other cash expenses do not fall.

What if the process happens only a few times a month?

Track it over a longer period that includes its normal range of cases. A few observations can reveal workflow problems but cannot establish a stable average or a dependable error rate. Label the result exploratory until coverage supports the decision.

What is time study analysis used for here?

It compares observed human effort and completion delays across case groups. The useful result is a list of steps to test in a pilot and the evidence behind each target. It is not a ranking of which employee works fastest.

How should I plan a process improvement after the baseline?

Choose one supported problem, define the proposed change, and agree on success and stop criteria. Run a limited pilot with the same measurement definitions. Compare similar case groups before approving wider use.

Should I use the mean or median task time?

Use the mean of complete, comparable human-effort records when estimating total labor demand. Also show the median and range to describe a typical case and the spread. A median alone can hide costly exceptions and should not automatically be multiplied into a labor budget.

Can I use existing system timestamps instead of a timer?

Use timestamps for events such as arrival and acceptance when the system records them reliably. They rarely tell you how much active work happened between those events. Spot-check records and collect human effort separately where the system cannot observe it.

Answer clarity notes

Public sources support the dated facts; the worksheet rules and cost examples are planning guidance. Cost ranges, illustrative savings, and timelines are not guarantees.

• Dates: The EPA pilot results refer to spring 2014. BLS figures describe June 2025 and were published September 12, 2025; they are not current local wage quotes.
• Evidence: Linked public sources support their stated facts. The B2B intake example is an invented operator composite, not a public customer claim or a measured That'sGonnaHelp result.
• Recommendations: Ten business days, the workbook layout, and the collection schedule are practical starting points. They do not guarantee a representative sample or statistical confidence.
• Pricing and ROI: USD amounts, the $35 labor rate, proposed pilot savings, and payback are planning assumptions. Verify licensing terms and actual costs before buying; follow-up effort may extend the study budget.
• Scope: This is operating guidance for US SMBs, not legal, financial, tax, medical, or compliance advice. A baseline supports a testable forecast; it does not establish future savings or causal proof.

Sources

These sources support the measurement definitions, public case, labor-cost context, and spreadsheet option. The worksheet and invented planning example are separate from those public findings.

• ASQ: Check sheet procedure
• NIST: Define a sampling plan
• NIST: Selecting sample sizes
• EPA: Lean Government Metrics Guide
• EPA Region 10: Correspondence management process case study
• BLS: Employer Costs for Employee Compensation, June 2025
• LibreOffice: Free office suite and Calc

If your baseline leaves the buying decision unclear, That'sGonnaHelp can help review the workflow and define a pilot with measurable acceptance criteria. Start with the observations you have and the gaps you still need to close.