Requirements Gathering Techniques

Observation & Job Shadowing

18 min Lesson 4 of 10

Observation & Job Shadowing

Interviews and questionnaires capture what people say they do. Observation captures what they actually do. That gap — sometimes narrow, often enormous — is one of the most productive seams a systems analyst can mine. When a warehouse supervisor explains her receiving process in a meeting room, she describes the ideal flow. When you stand beside her on the loading dock at 7 a.m., you discover the workaround spreadsheet she has been running for three years because the ERP system cannot handle partial shipments.

Observation and job shadowing are ethnographic elicitation techniques: they borrow from social science the idea that the best way to understand a practice is to witness it in its natural setting. For business analysts, that setting is the shop floor, the call centre, the clinic reception desk, or wherever the real work happens.

Why Observation Finds What Other Techniques Miss

People are not unreliable sources — they simply cannot describe everything they do, because much expert knowledge is tacit: embedded in muscle memory, professional habit, and contextual judgment. An experienced radiographer at a hospital might not mention that she always cross-references two screens simultaneously, because it is as automatic as breathing. Yet a system that forces a single-screen workflow would break her productivity and, potentially, patient safety.

Observation is the technique of choice when:

The current process is complex, rapid, or involves physical artefacts (paper forms, physical goods, multiple monitors).
Users have difficulty articulating their own workflow (common with long-tenured staff).
You suspect a significant difference between the documented process and the real one.
You need precise data on task duration, error frequency, or exception rates.
You are designing a system that must fit seamlessly into an existing physical or social environment.

Tacit knowledge: Knowledge that experts hold but cannot easily express in words. Observation externalises tacit knowledge by making it visible. It is the only reliable way to capture workarounds, informal coordination, and exception handling that never appear in process documentation.

Types of Observation

Not all observation looks the same. The analyst must choose a stance along two axes: how visible they are to workers, and how much they participate in the work.

Passive observation — The analyst watches quietly without interrupting the workflow. Workers are aware of being observed but the analyst asks no questions during the session. Ideal for high-volume, repetitive tasks (a call-centre agent handling 40 calls per shift) where interruptions would distort timing data.
Active observation (job shadowing) — The analyst follows one worker through their full shift, asking clarifying questions in real time. Deeper understanding of decision logic, but more disruptive and time-consuming.
Participant observation — The analyst actually performs the job alongside workers, usually for a short period. Generates deep empathy and uncovers ergonomic and cognitive demands that watching alone cannot reveal. Requires care to avoid disrupting production.
Video / screen recording — Software tools capture exactly what a user does on screen (mouse path, keystrokes, application switches). Objective, replayable, and excellent for spotting UI inefficiencies or illegal workarounds in an ERP. Requires privacy policy compliance.

Four observation modes mapped by analyst participation and workflow disruption — choose based on task type and information goal.

The Hawthorne Effect — The Observer Changes What Is Observed

In the 1920s and 1930s, researchers studying productivity at Western Electric's Hawthorne plant noticed something strange: workers became more productive simply because they were being watched — regardless of what the researchers actually changed. This became known as the Hawthorne Effect: people modify their behaviour when they know they are under observation.

For business analysts, the Hawthorne Effect is a constant threat to the validity of observation data. A customer service agent who normally copies-and-pastes data between two systems may stop doing so while watched, following the official process instead. An accounts-payable clerk who usually skips an approval step on small invoices will dutifully seek approval during your observation session. The result is that you record the aspirational process, not the actual one — exactly what you were trying to avoid.

Strategies to reduce the Hawthorne Effect include:

Extended presence: Spend multiple days, not just a single session. People cannot sustain unusual behaviour indefinitely; within a day or two most workers revert to their natural patterns.
Frame it as process learning, not evaluation: Emphasise to workers that you are studying the system, not judging their performance. "I want to understand how the system makes your job harder" disarms defensiveness.
Triangulate with other data: Compare what you see with system logs, error reports, and help desk tickets. If the ERP shows 200 data-entry corrections per day but you observe zero, the gap reveals that corrections are being suppressed during your visit.
Anonymous screen recording: Aggregated session data with personal identifiers removed captures true usage patterns without making individuals self-conscious.

Never position observation as a performance audit. If workers believe you are measuring their productivity or looking for grounds to discipline them, trust evaporates and you will see only the performance they want you to see. Observation must be positioned — and genuinely conducted — as a process improvement exercise.

Ethnographic Observation Techniques in Practice

Ethnography, borrowed from anthropology, treats the workplace as a culture to be understood on its own terms. Ethnographic analysts use several specific practices:

Field notes: Handwritten or digital notes taken during observation. Record what happens (events), when (timestamps), who is involved, and what artefacts are used. Avoid interpretation during note-taking; capture raw facts first.
Artefact collection: Gather physical or digital specimens of what workers use — the sticky note grid on the monitor, the personal Excel tracker, the printed cheat sheet. These artefacts reveal gaps the official system leaves unfilled.
Think-aloud protocol: Ask the worker to narrate their thoughts as they work. "I'm checking this field because sometimes the supplier enters the wrong code — we get rejections if I don't." That sentence captures a business rule that no requirements document contains.
Exception spotting: Deliberately observe at peak times (Monday morning order rush, month-end closing, the arrival of a large delivery) when exceptions surface. Normal hours show the happy path; peak hours show reality.

A structured observation session moves through four phases; the Hawthorne risk is highest during the Observe phase and must be mitigated upfront in Prepare.

Conducting a Job Shadowing Session — A Worked Example

Imagine you are the lead analyst on a project to replace a logistics firm's manual shipment booking process with an online portal. You have already conducted stakeholder interviews; now you want to observe the booking clerks at work.

Before the session: Brief the operations manager. Confirm the purpose ("I want to learn how the booking process works so the new system fits your team's real workflow"). Obtain written consent from the clerks who will be shadowed. Schedule two or three sessions across different days and shift times.
During the session: Arrive before the shift starts. Sit to the side of the clerk, never between them and their work. Take timestamped field notes. When the clerk switches applications, note which systems are open simultaneously. When something unexpected happens — a carrier API fails, a customer calls to change a shipment mid-booking — resist the urge to interrupt; observe how the exception is handled, then ask about it afterwards.
Sample finding: At 09:14, a partial shipment arrives. The ERP rejects it because it expects complete manifests. The clerk opens a personal spreadsheet, records the partial, emails the warehouse supervisor, then manually creates a dummy booking in the ERP to "park" the shipment. This entire workaround takes 11 minutes and happens roughly eight times per shift — that is nearly 90 minutes per day of invisible rework.
After the session: Conduct a short debrief (15–20 minutes) with the clerk. "I noticed you used that spreadsheet several times — can you walk me through why?" This turns raw observation into articulated requirements: "The system must support partial manifests and allow bookings to be updated incrementally without rejection."

The 11-minute workaround rule: Any workaround that takes more than a few minutes and happens multiple times per shift is a requirement hiding in plain sight. Document every workaround with its frequency and time cost — the business case for fixing it practically writes itself.

Strengths and Limitations

Observation offers rich, contextual, and honest data that no other technique consistently produces. But it comes with real costs and constraints:

Time-intensive: A single half-day session per role is a minimum; multiple sessions over several days are ideal. This is far more expensive in analyst time than a questionnaire sent to 50 people.
Observer effect (Hawthorne): Workers may alter their behaviour; mitigate but cannot fully eliminate.
Scope is narrow: You see one person, one shift, one location. Rare events (month-end close, system outages) may not occur during your observation window.
Consent and privacy: Workers must agree to be observed. In some industries, union agreements or privacy law restrict what can be recorded and how data is stored.
Analyst interpretation bias: What you notice and record is shaped by your existing mental model. Bring a co-analyst or rotate observers to counter this.

When to Combine Observation With Other Techniques

Observation is most powerful as part of a multi-technique strategy. In a clinic booking system project, the typical combination looks like this:

Interviews first — to understand the big picture, roles, and pain points from management.
Observation next — to see actual work and surface tacit knowledge and workarounds.
Document analysis in parallel — to compare what the process manual says with what you observed.
Workshops afterwards — to validate findings with the whole team and prioritise requirements collaboratively.

Golden rule: Always follow an observation session with at least a brief interview with the observed worker. Observation tells you what happened; the follow-up interview tells you why. Together they produce requirements that are both accurate and traceable.

Summary

Observation captures what people actually do — not what they say they do — revealing tacit knowledge and workarounds.
Four modes: passive observation, job shadowing, participant observation, and screen/video recording.
The Hawthorne Effect is the tendency for workers to change behaviour when observed; counter it with extended presence, transparent framing, and data triangulation.
Ethnographic techniques (field notes, artefact collection, think-aloud, exception spotting) extract the deepest insight.
Every observation session should follow the Prepare → Observe → Debrief → Analyse lifecycle.
Workarounds observed in the field are requirements hiding in plain sight — document them with frequency and time cost.
Observation is resource-intensive; combine it strategically with interviews, document analysis, and workshops.