公開日:
Work Analysis
What Is Work Analysis in Manufacturing? Methods, Steps, and Examples

Work analysis is the systematic observation of how a task is performed. In manufacturing, teams break a job into work elements and record movements, elapsed time, and waiting. The goal is to identify what causes delays or variation and decide which change to test on the shop floor.
This guide uses an automotive component assembly example to explain how to choose a method, record the current process, and measure the effect of an improvement.
- What should a work analysis reveal?
- Choose a method based on the question
- How to conduct a work analysis in six steps
- 1. Turn the concern into a measurable question
- 2. Define the start, finish, and work elements
- 3. Record routine work and exceptions
- 4. Check long intervals against the evidence
- 5. Choose one improvement and check its conditions
- 6. Measure again and update the standard
- What to record before observing
- Example: Reduce time spent looking for tools
- What AI can help with
- Frequently asked questions
- How does work analysis differ from labor tracking?
- How many cycles should I observe?
- Summary
- Use Ollo Factory to analyze manufacturing work from video
- Watch Ollo Factory’s Work Analysis in Action
What should a work analysis reveal?
Look beyond how fast the operator works. Material supply, tool placement, automatic machine cycles, and inspection procedures all affect the time a job takes. The most useful observations connect the delay to a specific event and its operating conditions.
To understand where time goes, record work element durations or the proportion of observations in each activity.
To reduce travel, draw the route on a layout and record distances or trips.
To understand differences in method, compare videos of the same task under comparable conditions.
Choose a method based on the question
Question | Method | What to look for |
|---|---|---|
Which step takes the most time in a cycle? | Element durations and the events behind long observations. | |
How is time distributed across activities during a shift? | The estimated share of time spent in defined activity categories. | |
When do an operator and a machine wait for one another? | The sequence and overlap of manual work, walking, and machine time. | |
Are workloads balanced across stations? | Work content by station or operator compared with takt time. | |
Where do parts travel or wait? | Process analysis and a travel diagram | Process steps, transport distance, and points where work piles up. |
Where do experienced and new operators use different methods? | Side-by-side video comparison | Differences in sequence, movement, and inspection or decision points. |
Match the observation window to the question. Sampling a shift can estimate how often an activity occurs. Timing a cycle gives detail about individual steps. To investigate both, use complementary observations.
How to conduct a work analysis in six steps
1. Turn the concern into a measurable question
For a slow fastening station, ask which part of the cycle from part verification through final inspection is associated with exceeding takt time. Record the part number, required pace, shift or time window, and conditions you will compare.
2. Define the start, finish, and work elements
Use visible events to mark the boundaries. In component assembly, the work elements might be picking a part, seating it in a fixture, fastening it, checking the inspection result, and unloading it. Share a reference recording and a coding rule so different observers use comparable boundaries.
3. Record routine work and exceptions
Observe multiple cycles. Flag time spent searching for parts, waiting for the machine, handling a defect, or replenishing materials. Include the circumstances of each event. A single fast cycle can miss recurring delays.
4. Check long intervals against the evidence
If the interval before fastening is long, check whether similar parts are hard to distinguish, components are missing, or the fixture makes seating difficult. Use video, shop-floor observations, and operator input to distinguish between layout, equipment, materials, method, and training issues.
5. Choose one improvement and check its conditions
If a parts bin is difficult to identify, test clearer labels. Check selection accuracy, ease of picking, and replenishment. If the proposed change affects sequence, verify the approved quality and safety conditions with the responsible team before trialing it.
6. Measure again and update the standard
Keep the part, observation boundaries, and categories consistent before and after the change. Compare both typical times and long cycles, along with defects and workload. Document the successful method in the relevant instructions, layout, and training materials.
What to record before observing
Field | Automotive assembly example |
|---|---|
Question | What causes delays before fastening? |
Conditions | Use the same part number, fixture, and parts layout. Record any differences separately. |
Boundaries | Start at part verification and end when the inspection result is confirmed and the unit is removed. |
Categories | Verification, loading, fastening, inspection, unloading, searching, and waiting. |
Working hypothesis | Part identification and seating in the fixture may be causing delays. |
Supporting evidence | Bin labels, fixture condition, replenishment times, and operator observations. |
Follow-up measures | Time under the same conditions, errors, workload, and completion of required checks. |
Example: Reduce time spent looking for tools
Observe. Record the search time and where the tool was stored or returned.
Test. Assign an accessible storage position and a return rule. Check that operators still select the correct tool.
Recheck. Compare search frequency and duration for the same part and measurement window. Confirm that required checks still happen.
This example illustrates the method. The results of a real trial depend on the process and the conditions observed.
What AI can help with
Video analysis software can segment work, classify activities, aggregate time, and suggest improvement opportunities. The team still sets the work boundaries, category definitions, and acceptance criteria. For causes that occur off camera or are obscured by equipment, use additional evidence from the floor.
Frequently asked questions
How does work analysis differ from labor tracking?
Labor tracking shows where people’s hours were booked against jobs or products. Work analysis examines how the time was spent and which conditions could be improved. A labor report can identify a process to investigate in greater detail.
How many cycles should I observe?
Choose a sample that reflects the variation you need to understand. Consider product variants, shifts, replenishment, and exceptions. If observations miss an important operating condition, collect more data for that condition.
Summary
Define the question and choose a method that can answer it.
Use consistent boundaries and document the conditions alongside the observed events.
Test a single change, then compare time, quality, and workload before revising the standard.
Use Ollo Factory to analyze manufacturing work from video
Ollo Factory can break a work video into elements, group them using the shop floor’s activity definitions, and summarize the time spent in each group. Teams can review the source footage to investigate a long interval or verify a suggested improvement before testing it.
Keep the analysis definitions consistent across recordings and have the people responsible for the process review the results against equipment, quality, and safety requirements.
Watch Ollo Factory’s Work Analysis in Action
この記事の執筆者
株式会社Ollo
事業開発部リーダー マーケティング担当 Yoshida
製造業特化型の作業分析AI「Ollo Factory」を開発・提供する株式会社Ollo(東京大学 松尾研究室発のAIスタートアップ)で、事業開発とマーケティングを担当。
大学時代は工学部で設計開発を学び、新卒で大手自動車メーカーに入社。新型EVの開発グループで、設計から工場での生産まで幅広くプロジェクトに携わり、研修の一環として、実際に自動車製造工場でのライン作業も経験。その後、製造業領域のスタートアップで営業・マーケティングに従事。
現在は、製造現場でのAI活用を促進し、よりワクワクする働き方を実現できるよう、「Ollo Factory」の販売・マーケティングに取り組んでいる。
一覧に戻る
関連記事
Ollo Factoryについて、さらに詳しく。

まずは資料請求
・ 導入事例と効果
・ 主な機能
・ 活用の流れ

