Observation and field research: seeing how work and customers really behave (Research and Evidence for Better Decisions) | READOS

Ask people how they do their work and they will usually describe the official process, the way it is meant to happen or the way they did it last time something went well. Watch them, and a different picture often appears: workarounds for a missing tool, a spreadsheet kept on the side because the system is too slow, three trips to the store for parts that could have been kitted, a customer who searches the website twice before giving up and phoning. None of this is dishonesty. Routine actions become invisible to the people who perform them, and the gap between what people say and what they do is one of the most consistent findings in research.

Observation closes that gap. Watching work, customers and processes in their real settings, often called field research, reveals how things actually happen, where time and effort are lost, what frustrates people and what they value. It underpins lean improvement, product design, service design and process redesign. It can be as simple as spending an hour on the shop floor with a notebook or as structured as a statistically designed work-sampling study.

Like any research method, observation has its own biases and risks. People behave differently when watched. Observers see what they expect to see. Notes written later drift from what happened. Photographs and video raise privacy concerns. This article explains the main forms of observation, how to plan and conduct it, how to record observations reliably, how to reduce observer effects, how to handle ethics and safety, and how to turn observations into evidence. It is part of a series on research and evidence for business decisions.

Why observe

Observation is particularly valuable when:

processes are routine, so people cannot easily describe them;
the question is about what happens, not what people think;
problems are hidden in workarounds, waiting, rework or searching;
users struggle with products, websites, forms or equipment; and
interview or survey findings need checking against actual behaviour.

It is less suitable for understanding reasons, which interviews do better, or for measuring how common something is across a large population, which surveys and records do better. Observation is often most powerful combined with brief conversations: watching a task and then asking why it was done that way.

Forms of observation

Unstructured observation

The observer watches and records what seems relevant, without a fixed checklist. It suits early exploration, when the important issues are not yet known. Field notes record events, sequences, quotes and the observer’s questions.

Structured observation

The observer records specific, predefined behaviours or events using a checklist or coding sheet: counts of interruptions, time spent on each activity, number of steps, errors or near misses. Structured observation produces data that can be counted and compared.

Participant and non-participant observation

In participant observation, the researcher takes part in the work, such as working a shift on the line or handling customer calls. It gives deep understanding but can blur the line between doing and observing. In non-participant observation, the researcher watches without taking part.

Overt and covert observation

Overt observation is done openly, with the people observed aware of it and its purpose. In workplaces, observation should generally be overt: covert observation of staff raises serious ethical, legal and trust issues and should not be used without careful justification and advice. Some forms of customer observation in public settings, such as counting foot traffic, are less intrusive, but privacy obligations still apply when individuals can be identified.

Gemba walks

In lean practice, the gemba is the place where value is created, such as the shop floor, warehouse or service counter. A gemba walk is a deliberate visit by managers and improvement teams to observe work, ask respectful questions and understand problems first-hand rather than through reports. Effective walks have a focus, such as material flow or safety, involve listening rather than instructing and lead to follow-up actions. The small, frequent improvement on the shop floor article explains how gemba walks support continuous improvement.

Contextual inquiry

Contextual inquiry combines observation and interviewing. The researcher watches a person doing their real work in their real setting and asks questions as they go, in a relationship sometimes compared with that of an apprentice learning from a master. It is widely used in designing software, equipment and services, because it reveals the context, tools, interruptions and workarounds that shape how people actually use products.

Measuring work: time studies and work sampling

Time studies time specific tasks with a stopwatch or video to establish standard times. They require care: the person timed should be trained and working at a normal pace, enough cycles should be timed to capture variation and allowances should be made for rest and unavoidable delays.

Work sampling takes many random snapshots of what people or machines are doing, recording the activity at each instant: working, walking, waiting, searching, setting up, idle. The proportion of observations in each category estimates the proportion of time spent on it. With several hundred observations, the estimate is reasonably precise: for an activity observed 31% of the time in 400 random observations, the 95% margin of error is about ±4.5 percentage points. Work sampling is less intrusive than continuous timing and suits activities that vary a lot. Work sampling is also a practical way to estimate how much of a machine’s available time is spent running, set up, waiting or broken down.

Observing customers

Customer observation takes several forms:

Site visits: watching how customers use products in their own premises.
Usability testing: asking people to complete realistic tasks with a product, website or form while thinking aloud, and noting where they hesitate, err or give up. A handful of participants often reveals most of the serious problems.
In-store or counter observation: watching how customers browse, ask questions and decide.
Mystery shopping: trained people posing as customers to experience the service and report against defined criteria.
Digital analytics: observing behaviour on websites and apps through aggregated data on navigation, searches and abandonment.

The user experience for products, services and processes article describes how such observations improve design.

Field research in markets

Field research also means getting out into markets: visiting trade shows, competitor showrooms and retail outlets, walking construction or industrial sites, attending industry events and speaking with distributors and installers. Structured notes on what is observed, such as product ranges, prices, displays and how staff respond to questions, turn visits into comparable evidence rather than impressions.

From observations to process maps

Observations become most useful when they are organised. A process map or flow chart drawn from what was observed, rather than from the official procedure, shows the real sequence of steps, decisions, handoffs, waits and loops. Spaghetti diagrams, which trace the path of a person or item across a floor plan, make wasted movement vivid. Mapping the current state with the people who do the work, and marking where problems were seen, builds shared understanding and is a natural starting point for improvement.

Observing safety

Observation is also a safety tool. Watching how tasks are actually performed reveals hazards that procedures and risk assessments miss: guards bypassed to clear jams, awkward lifts because a trolley is missing, shortcuts through forklift areas. Safety observations work best when they are framed as learning about the work rather than catching people out, when observers thank workers and discuss what they saw and when findings lead to changes in equipment, layout or procedures rather than only reminders to follow the rules.

Planning observation

A short plan helps:

Purpose and questions: what the observation should reveal.
What and whom to observe: tasks, roles, locations and customer groups.
When: times, shifts, days and seasons, chosen to represent normal variation rather than only convenient moments.
Method: unstructured notes, checklists, timing, work sampling or video.
Permissions, ethics and safety: consent, briefing, site rules and personal protective equipment.
Recording and analysis: formats, storage and how findings will be combined.

The observer effect

People change their behaviour when they know they are being observed, often working faster, more carefully or more by the book. This observer effect, sometimes called the Hawthorne effect, can be reduced by spending time in the setting until people become used to the observer, observing over longer periods, explaining clearly that the focus is on the process rather than individual performance, and comparing observations with records that are not affected by observation.

Observer bias

Observers also bring expectations, noticing what confirms their views and missing what does not. Using clear definitions for structured categories, having two observers code the same events and compare results, separating descriptive notes from interpretation and reviewing observations with the people observed all reduce bias.

Recording observations

Field notes should be written during or immediately after observation. Good notes separate:

descriptive notes: what happened, in factual terms, with times and sequences;
quotes: what people said, in their words; and
reflective notes: the observer’s interpretations, questions and ideas.

Photographs, sketches and process maps capture layouts and flows. Video is powerful for analysing tasks in detail but requires consent and careful handling, because it records identifiable people.

Ethics, privacy and safety

People observed at work should be told what is being observed, why, how information will be used and how they will be protected. Findings should be used to improve processes, not to discipline individuals, unless that purpose has been made clear and appropriate processes followed. Images and recordings of identifiable people are personal information and must be stored and used accordingly. Observers must follow site safety rules, complete inductions, wear required protective equipment and never put themselves or others at risk to get a better view.

A worked example

This is an illustrative example. The business and figures are hypothetical.

A parts distributor’s warehouse struggles to dispatch orders on time, and staff ask for more pickers. Before hiring, the operations manager conducts a two-week work-sampling study, recording at 400 random moments what each picker is doing. Staff are briefed beforehand that the study is about the layout and process, not individual performance.

The results show pickers spend about 31% of their time walking, 14% searching for stock in the wrong location, 9% waiting for the single packing bench and only about 38% actually picking. Unstructured observation adds detail: the fastest-moving items are stored at the far end of the building, and pickers make extra trips because the order sheets are not sorted by location.

The warehouse re-slots the 200 fastest-moving items near dispatch, sorts pick lists by location and adds a second packing bench. A repeat study a month later shows walking down to about 18% and searching to about 5%, and on-time dispatch improves without extra staff. The lean warehousing and material flow article explains the slotting principles behind the changes.

How this applies to a small Australian business

Watch the work before redesigning a process or adding resources.
Combine observation with brief questions to understand why.
Use work sampling to see where time really goes.
Test products and forms with real users completing real tasks.
Observe at representative times, not just convenient ones.
Tell staff what is being observed and why, and use findings to improve processes.
Separate description from interpretation in notes.
Map the real process with the people who do the work.
Use safety observations to learn about the work, not to catch people out.
Follow site safety rules while observing.

Common mistakes

Relying only on what people say about routine work.
Observing only at quiet or convenient times.
Using observations to blame individuals.
Covert observation of staff.
Writing notes days later.
Mapping the official procedure instead of what actually happens.
Seeing only what confirms expectations.

Frequently asked questions

How long should observation last? Long enough to see normal variation and for people to become used to the observer. For many processes, several sessions across different days and shifts work better than one long session.

Is filming staff acceptable? Only with clear notice, a legitimate purpose, secure handling and compliance with privacy and workplace surveillance laws, which vary between states and territories. Seek advice where in doubt.

How many users are needed for usability testing? A small number, often around five per round, typically reveals the most serious problems. Several rounds with improvements in between are more effective than one large test.

What is the difference between a gemba walk and an inspection? A gemba walk seeks to understand how work happens and what gets in the way; an inspection checks compliance. Mixing the two discourages people from showing real problems.

Questions to ask

Which of our processes have we never watched from start to finish?
Where might routine workarounds be hiding wasted time or risk?
Have we watched customers using our products, website or forms?
How do we make sure observation is fair, safe and respectful?
Do our process maps show how work really happens, including waits and workarounds?
What hazards would we notice if we watched our riskiest tasks for an hour?
How do we turn observations into measured improvements?

Bringing it together

Observation shows what people actually do, closing the gap between official processes and real practice. Unstructured and structured observation, gemba walks, contextual inquiry, time studies, work sampling, usability testing, site visits and market field research each suit different questions. Planning, representative timing, careful notes, attention to observer effects and bias, and respect for ethics, privacy and safety turn watching into reliable evidence for improvement.

Source: KEVOS editorial notes. Examples and figures in this article are illustrations. This article is general information and does not constitute legal advice on privacy or workplace surveillance.