Observation (Ethnographic / Contextual Inquiry)

Purpose

Observe users in their natural context performing actual tasks, workflows, or experiencing the validated need in real-time, capturing behaviors, environmental factors, workarounds, pain points, and artifacts that users may not articulate in interviews. Observation reveals the gap between what users say they do (interview data) and what they actually do (observed reality), uncovering tacit knowledge, unconscious habits, and contextual factors invisible to users themselves.

Ethnographic observation (passive watching) and contextual inquiry (observation + concurrent questioning) provide complementary data to interviews—grounding verbal reports in observable reality and revealing environmental, social, and workflow factors shaping user behavior.

When to Use

Use observation when:

• Need involves complex workflows, tasks, or behaviors better observed than described
• Users may not accurately recall or articulate their actual behavior (tacit knowledge, habitual actions)
• Context matters—physical environment, tools, interruptions, collaboration significantly shape the experience
• Seeking to understand workarounds, hacks, or adaptations users have invented
• Validating interview findings—checking if stated behavior matches observed behavior
• Early exploration—don't yet know what questions to ask, observation generates hypotheses

Do NOT use (or deprioritize) when:

• Need is primarily emotional, aspirational, or hypothetical (future behavior)—interviews better
• Observation would be invasive, unethical, or unsafe (sensitive contexts, privacy concerns)
• User behavior is infrequent or unpredictable—cannot realistically observe need arising (e.g., crisis events)
• Time/budget constraints severe—observation more resource-intensive than interviews
• Remote/distributed users with no shared physical context (though screen-share observation possible)

Sample Size and Duration

Number of observations: 6-12 sessions (typical for A1.2)

• Minimum: 3-5 (if combined with interviews—triangulation)
• Saturation: Continue until patterns repeat, no new insights (often 8-10 sessions)
• Variability: Observe across user types, contexts (morning vs. afternoon, busy vs. slow periods, novice vs. expert)

Session duration: 1-3 hours per observation

• Task-dependent: Observe full task cycle (e.g., complete forecast prep workflow)
• Typical breakdown: itemize
• 10 min: Setup, consent, rapport-building
• 60-120 min: Observation (passive ethnographic and/or contextual inquiry)
• 15-30 min: Post-observation interview (clarifying questions, artifact review) itemize

Analysis time: 2-3 hours per observation session

• 30-60 min: Transcribe/organize notes
• 30-60 min: Tag themes, extract quotes
• 30-60 min: Cross-session pattern analysis (after every 3-4 sessions)

Prerequisites

• Access to users in context: Permission to observe in workplace, home, or relevant environment
• Observation protocol: Structured guide (what to observe, how to record notes)
• Recording tools: Notebook + pen (minimum), audio recorder (if permitted), camera/video (if permitted and ethical)
• Informed consent: Users aware of observation purpose, data use, confidentiality
• Time allocation: 1-3 hours per observation session (depends on task duration)
• Researcher training: Skills in non-intrusive observation, concurrent note-taking, contextual questioning

Complete Procedure

Pre-Observation Preparation

Step 1: Define Observation Focus and Protocol (1-2 hours)

Create observation protocol specifying:

1. Research questions: What are you trying to learn from observation? itemize
2. Example: “How do sales managers assess deal health when preparing weekly forecasts?” itemize
3. Context/setting: Where will observation occur? itemize
4. Example: Manager's desk during Monday morning forecast prep (9-11 AM) itemize
5. Observation targets: What specifically to observe?

Observation target framework (AEIOU):

Step 2: Recruit Participants and Schedule Sessions

• Sample size: 6-12 observation sessions (typical for A1.2)
• Timing: Observe when need/task naturally occurs (not artificial demonstration) itemize
• Good: Observe Monday morning forecast prep (when managers actually do it)
• Bad: Ask manager to “show me how you forecast” on random Thursday (artificial, may not reflect real behavior) itemize
• Consent: Explain purpose, recording methods, data confidentiality; obtain written/verbal consent

During Observation Session

Observation Mode 1: Passive Ethnographic Observation

Approach: Researcher observes silently without interruption—“fly on the wall.”

When to use:

• Want to minimize observer effect (user behavior change due to being watched)
• Task is time-sensitive or flow-dependent (interruption would distort)
• Initial observation—building understanding before asking questions

Technique:

• Position: Sit/stand where you can observe but not intrude (side angle, not directly behind/over shoulder)
• Note-taking: Write continuously—capture actions, artifacts, environment, quotes (if user talks aloud), timestamps
• Objectivity: Describe what you see, not what you interpret itemize
• Good: “Manager furrows brow, sighs, taps pen on desk while staring at Excel”
• Bad: “Manager is frustrated” (interpretation—save for analysis) itemize
• Questions: Save for after task completion (unless user spontaneously explains)

Observation Mode 2: Contextual Inquiry (Active Observation + Questioning)

Approach: Researcher observes and asks questions concurrently—“master-apprentice” model (user is master showing apprentice how work is done).

When to use:

• Task is not highly time-sensitive or flow-dependent
• Need to understand why user does something, not just what they do
• User is comfortable with interruption and enjoys explaining their work

Questioning types:

1. Clarification questions: “What are you doing now?” “What is that column?”
2. Rationale questions: “Why did you choose X?” “What made you decide to call the rep instead of emailing?”
3. Workaround questions: “Is this how the system is supposed to work, or is this a workaround you invented?”
4. Comparison questions: “How does this compare to how you did it last week/month/year?”
5. Artifact questions: “Can you show me that spreadsheet?” “Where did this sticky note come from?”

Timing of questions:

• During task: Brief, clarifying questions that don't disrupt flow
• After task completion: Deeper questions about rationale, pain points, alternatives

Field Note Structure

Real-time notes (during observation):

Use shorthand template:

verbatim TIME | ACTION | ARTIFACT/TOOL | QUOTE/COMMENT ——+———————-+—————+—————— 9:03 | Opens CRM, clicks | Salesforce | (mutters) "Ugh, | Reports tab | | so slow today" ——+———————-+—————+—————— 9:05 | Exports opp list | Excel export | Clicks "Custom | to Excel, opens in | template | Fields" - has | separate template | "Weekly FC" | 8 extra columns ——+———————-+—————+—————— 9:08 | Calls rep (Mike) | Phone | "Hey Mike, Stage | to verify Stage 3 | | 3 deal - is legal | deal status | | done?" [Rep: "Not | | | yet, maybe Friday"] ——+———————-+—————+—————— 9:12 | Updates Excel col | Excel "Legal | Writes "P" for | "Legal Status" | Status" | "Pending" in cell ——+———————-+—————+—————— verbatim

Post-observation notes (immediately after session):

• Contextual summary: Setting, participants, duration, task observed
• Behavioral themes: Patterns noticed (e.g., “Manager called reps 4 times—doesn't trust CRM data”)
• Pain points: Observed frustrations, errors, workarounds
• Artifacts captured: Photos of workspace, screenshots of tools, copies of documents (if permitted)
• Follow-up questions: Questions for next interview or observation
• Initial interpretations: Hypotheses, connections to research questions (clearly labeled as interpretation)

Post-Observation Analysis

Step 1: Transcribe and Organize Notes (30-60 min per session)

• Type up handwritten field notes
• Organize by AEIOU framework (Activities, Environment, Interactions, Objects, Users)
• Tag themes (workflow, workaround, pain point, collaboration, tool use, etc.)

Step 2: Identify Patterns Across Sessions

After 6-12 observations, conduct cross-session analysis:

1. Workflow mapping: Synthesize observed steps into typical workflow (journey map input)
2. Workaround inventory: List all workarounds, hacks, manual processes observed
3. Artifact catalog: Document tools, templates, documents users created/rely on
4. Pain point frequency: Count how many users exhibited same frustration/error
5. Environmental factors: Identify context factors shaping behavior (interruptions, tool constraints, social dynamics)

Step 3: Compare Observation Data to Interview Data

Say-Do Gap Analysis:

Gap interpretation:

• Gaps reveal tacit knowledge (users unaware of actual behavior), social desirability bias (report ideal behavior, not real), or memory errors (misremember frequency/duration)
• Observation data often more reliable for behavioral facts; interview data better for motivations/emotions

Quality Criteria

Excellent observation data demonstrates:

1. Contextual richness: Captures environment, tools, artifacts, social interactions—not just isolated task steps
2. Behavioral specificity: Describes observable actions (“clicks Export, opens Excel”), not interpretations (“is frustrated”)
3. Temporal structure: Notes include timestamps or sequence markers—can reconstruct workflow chronologically
4. Workaround identification: Documents hacks, manual processes, adaptations users invented
5. Artifact collection: Photos of workspace, screenshots of tools, copies of documents (where permitted)
6. Say-do comparison: Observation findings compared to interview data—gaps documented
7. Pattern validation: Behaviors observed across multiple users (not just individual idiosyncrasies)
8. Researcher reflexivity: Notes distinguish observation (“user did X”) from interpretation (“I think this means Y”)

Ethnographic Observation vs. Contextual Inquiry

Hybrid approach (recommended):

1. Start with ethnographic observation (15-30 min)—passive, natural behavior
2. Transition to contextual inquiry—ask clarifying questions during less time-sensitive moments
3. End with retrospective interview (15-30 min)—deeper questions about what you observed

Challenges and Solutions

Challenge 1: Observer Effect (Hawthorne Effect)—User Changes Behavior When Watched

Symptoms:

• User performs task unusually carefully, slowly, or formally
• User explains every action (“Now I'm clicking X”)—artificial narration
• User skips workarounds or shortcuts they'd normally use

Solutions:

• Build rapport first: Spend 10-15 min chatting before observation—reduce anxiety
• Normalize observation: “I'm just here to learn how you work—please do exactly what you'd normally do, don't change anything for me”
• Longer sessions: After 30-60 min, users often forget you're there and revert to natural behavior
• Multiple sessions: First session may be artificial; second/third sessions more natural
• Blend in: Dress/act like you belong in the environment (don't stand out as outsider)

Challenge 2: Access and Logistics—Hard to Observe in Natural Context

Symptoms:

• Users work remotely (distributed, no shared physical space)
• Observation requires travel, scheduling complexity
• Privacy/security concerns (can't observe in secure facilities, with sensitive data)

Solutions:

• Screen-share observation: For digital work, use Zoom/Teams screen-share—observe software usage, workflows (though miss physical environment)
• Recorded task videos: Ask users to record themselves performing task (Loom, phone camera)—review asynchronously (though miss real-time questions)
• Diary studies with photos: Users document tasks with timestamped photos, notes—approximates observation (see Diary Studies method)
• Prioritize highest-value observations: If can only observe 3 users in-person, choose most representative/extreme cases

Challenge 3: Note-Taking Overload—Can't Capture Everything

Symptoms:

• Researcher misses key moments while writing notes
• Notes incomplete, illegible, or lack context

Solutions:

• Audio/video recording: If permitted, record session—take sparse notes during, detailed notes after while reviewing recording (but recordings time-intensive to review)
• Two researchers: One observes and asks questions, one takes detailed notes (doubles cost but much richer data)
• Structured note template: Pre-formatted table (Time | Action | Tool | Quote) faster than freeform notes
• Abbreviations: Develop shorthand (U = user, CR = CRM, FC = forecast, etc.)
• Focus on critical moments: Note everything during key task steps, less detail during routine/repetitive parts

Challenge 4: User Interruptions—Natural Context Includes Distractions

Symptoms:

• During observation, user receives phone call, coworker interrupts, meeting starts
• Task gets fragmented, hard to observe complete workflow

Solutions:

• Observe interruptions as data: Interruptions are part of real context—note frequency, type, impact on task
• Ask to resume: “When you're done with that call, could we pick up where you left off?”
• Schedule protected time: Ask user to block 2-hour window, minimize interruptions (though won't eliminate all)
• Multiple shorter sessions: 30-min sessions across 3 days may capture full workflow better than single 2-hour session

Tools and Materials

Essential:

• Notebook + pen (always works, no tech failures)
• Observation protocol/checklist (AEIOU framework, research questions)
• Informed consent form

Optional (if permitted):

• Audio recorder (captures user explanations, quotes)
• Camera/phone (photos of workspace, artifacts, tools)
• Video camera (full session recording—rich but time-intensive to analyze)
• Laptop for note-taking (if you type faster than you write)

Digital observation (remote):

• Zoom/Teams with screen-share and recording
• Loom/CloudApp (user self-records screen)
• Mobile screen recording apps (for mobile task observation)

Relationship to Other Methods

Observation complements:

• Interviews: Observation validates interview claims (say-do comparison), reveals tacit knowledge interviews miss
• Diary Studies: Diaries capture frequency/breadth (many instances over time); observation captures depth (single instance in detail)
• Journey Mapping: Observation provides step-by-step workflow data to populate journey map
• A6 Usability Testing: A1.2 observation establishes baseline behavior; A6 observation shows how prototype changes behavior

Observation provides input to:

• Journey Maps: Observed workflow steps → journey stages, touchpoints, pain points
• Empathy Maps: Observed behaviors, quotes, artifacts → See/Say/Do/Hear quadrants
• Affinity Diagramming: Observation notes become data points for thematic clustering
• A1.3 Root Cause Analysis: Observed workarounds reveal root causes (“Why did user invent this hack?”)

Example: Sales Forecast Preparation Observation

Context: Observing mid-market sales manager during Monday morning weekly forecast preparation (9-11 AM, manager's desk).

Key Observations (condensed field notes):

Activities:

• 9:03: Opens Salesforce CRM, clicks Reports → Opportunities closing this quarter
• 9:05: Exports opportunity list to Excel, opens in separate custom template (“Weekly FC Template”)
• 9:08-9:42: Series of 4 phone calls to sales reps, asking about deal status (legal done? budget approved? champion confirmed?)
• 9:45-10:30: Manually updates Excel columns based on rep calls (adds 8 columns not in CRM: Legal Status, Budget Approval, Champion Strength, Risk Level, etc.)
• 10:35: Stares at spreadsheet for 5 min, changes 3 probabilities based on “gut feel”
• 10:40: Copies final forecast numbers into email to VP

Environment:

• Dual monitors (CRM left, Excel right); phone within arm's reach
• Sticky notes on monitor bezel with deal acronyms and status codes
• Coffee cup, stressed body language (furrowed brow, sighing)
• 2 interruptions: Slack message (ignored), coworker dropped by (“Not now, doing forecast”)

Artifacts:

• Custom Excel template with 8 columns beyond CRM fields—manager created this 2 years ago, shares with other managers
• Handwritten notebook with deal notes from previous rep conversations
• Screenshot captured: Excel template showing health signal columns CRM lacks

Say-Do Gap:

• Interview: “Forecast takes 30 min”—Observed: 2 hours
• Interview: “I trust CRM data”—Observed: Called all 4 reps to verify, didn't trust CRM alone
• Interview: “Use data-driven assessment”—Observed: Final step was gut-feel adjustments

Key Insight: Manager has systematized 8 health signals (custom Excel columns) that CRM doesn't capture—this workaround reveals the gap driving forecast anxiety. Root cause hypothesis: CRM shows stage/probability, not health signals needed for confident assessment.