What is contextual inquiry?
Contextual inquiry, also called shadowing, is a research method where you watch someone do a real task in the place where they normally do it, and ask questions while they work. You sit next to them at the shop counter, in the kitchen or at their desk, or you watch their screen on a video call. The goal is to see what people actually do, including the steps they forgot to mention in an interview.
The method comes from Hugh Beyer and Karen Holtzblatt’s book Contextual Design. They suggest acting like an apprentice learning from a master: the customer is the expert in their own work, and you are there to learn how the work is done. People who give short answers in an interview will explain every step when they are teaching.
When should you use it?
- When your problem involves a process: bookkeeping, scheduling, packing orders, planning meals, handling returns, training new staff. Processes hide workarounds.
- After problem interviews, when people agree the task is annoying but can’t explain where the time goes.
- For B2B and small-business ideas above all. It also works for consumer tasks at home, if people are comfortable with a visitor or a screen share.
Skip it when the problem is a feeling or a rare event (choosing a wedding venue happens once). An interview about the last time it happened works better there.
How do you do it, step by step?
-
Pick a task and a time when it really happens. Ask your interviewees: “When do you next do [the task]? Could I watch, quietly, and ask a few questions?” Book the session at that moment: Monday morning order packing, Friday invoicing, the evening meal plan. A staged demo shows you the ideal version. You want the messy one.
-
Set the frame in 2 sentences. “Please do it exactly as you normally would, shortcuts and all. I’m here to learn how you do it, so pretend I’m a new assistant on day one.” Ask permission to take notes and photos of the workspace (never of customers or private data).
-
Watch first, ask second. For the first 10 minutes, mostly observe. Then ask about what you just saw:
- “What did you just do there?”
- “Why that spreadsheet and not the system?”
- “What would happen if you skipped that step?”
- “Where did that number come from?”
- “Is today a typical day? What’s different when it goes badly?”
-
Hunt for the workarounds. Write down every one of these, with the time it took:
- Sticky notes on screens or fridges
- Personal spreadsheets next to “official” tools
- Copy and paste between 2 or more apps
- Printouts, notebooks, photos of paper on a phone
- Messages to themselves (emails, WhatsApp “notes to self”)
- Steps done twice “just to be sure”
Each workaround is a place where existing tools fail them, and they already invented a fix.
-
Check your interpretation on the spot. When you think you see a pattern, say it out loud: “It looks like you keep a paper list because the system doesn’t show which returns are waiting. Is that right?” Let them correct you. Interpretation is one of the principles of Holtzblatt and Beyer’s method, and the corrections bring out the best details.
-
Time the steps. Use your phone stopwatch for the main task and the workarounds. “Handling one return: 11 minutes, of which 6 are copying details between 3 tabs.” Numbers make the problem concrete, for you and later for customers.
-
Write it up within the hour. One page per session: the steps in order, the workarounds with times, 3 to 5 quotes, photos with captions, and a list of surprises. After 4 to 6 sessions, mark the workarounds that appear on most pages.
What does a worked example look like?
Say Elif, a hypothetical founder, wants to help small online shops handle returns. In interviews, sellers said returns were “a bit of a hassle, nothing major”.
She watches 4 sellers: 2 in person on their packing day and 2 by screen share on Google Meet. What she sees is very different from what she heard:
- 3 of 4 keep a separate list of open returns: a paper notebook, a spreadsheet, and starred emails.
- Every return involves copying the order number, customer name and reason between the shop admin, email and the shipping site. Average time: 11 minutes per return.
- 2 sellers photograph damaged items on their phone, email the photos to themselves, then attach them to a refund note.
- One seller says “I just do them on Sunday so they don’t ruin my week.” That means customers wait up to 7 days for a refund, which shows up in her reviews.
Elif’s new problem statement: “Small sellers spend 10+ minutes per return copying details between 3 tools, so they batch returns weekly and annoy customers.” She can now test a narrow fix: one page that collects the return reason and photos from the customer and fills in the seller’s list automatically.
Which free tools help?
| Tool | What you use it for | Free plan |
|---|---|---|
| Google Meet | Watching a screen share of a remote task | Free with a Google account |
| Zoom | Screen share sessions, with recording if they agree | Free plan with time-limited meetings |
| Google Docs | One write-up per session with photos | Free |
| Phone camera and stopwatch | Photos of the workspace, timing each step | Built into your phone |
What mistakes should you avoid?
- Asking for a demo instead of the real task. People show you the tidy version. Book the moment when the work actually happens.
- Helping or suggesting during the session. The minute you say “have you tried…”, they start performing for you. Save ideas for afterwards.
- Taking photos of private data. Customer names, health or bank details stay out of your photos and notes. Photograph the desk and setup, and keep screens with personal data out of frame.
- Watching just one person. One seller’s notebook is a quirk. The same notebook at 3 sellers is a market.
- Ignoring the boring parts. The 4 minutes of copy and paste nobody complains about is where your product can live. Time everything.
How do you know it worked?
You come back with workarounds you never heard about in interviews, timed and photographed. Our rule of thumb: after 4 to 6 sessions, at least 1 workaround shows up in more than half of them and costs 30 minutes or more per week.
If people do the task smoothly and the tools they have already work, the problem is smaller than interviews suggested. That is a no-go for this task. Look one step earlier or later in their process, where the friction moved.
What should you do next?
Add what you saw to your customer journey map and the “does” box of your empathy map, then sort everything in interview synthesis. The fastest follow-up is a concierge test: do the workaround FOR a few customers by hand. The F/MS startup game helps women founders plan exactly that kind of next experiment with an AI co-founder.
FAQ
What is the difference between contextual inquiry and an interview?
An interview relies on what people remember and choose to tell you. Contextual inquiry shows you what they do, in the place they do it, including habits they no longer notice. Use interviews to find the problem and shadowing to see how it plays out.
Can I do contextual inquiry remotely?
Yes, for any task done on a computer or phone. Ask the person to share their screen on a video call while they do the real task. For physical tasks, ask them to hold the phone camera on their workspace, or to send photos of each step.
How many contextual inquiry sessions do I need?
Our rule of thumb is 4 to 6 sessions per customer type. Patterns in workarounds appear fast, because people in the same role tend to invent the same fixes.
Updated 26 October 2026 · Written by the Fe/male Switch team