Ethnographic research for product teams: watch people where the work happens

Ethnographic research is the study of people in their own setting: you go where they live or work, watch what they do, sometimes join in, and ask about it as it happens. For most product and market questions you do not need the months of fieldwork anthropologists do. A few half-day visits plus a short diary study answers many product questions, and plain interviews are enough when people can remember and explain the thing you want to know.

The reason to bother: people are bad at describing their own habits, and habits decide whether a product fits.

1. Ethnography means going to the people and staying long enough to see the ordinary

The method comes from anthropology. The open textbook Shared Voices: An Introduction to Cultural Anthropology describes how scholars such as Bronislaw Malinowski and Margaret Mead moved away from "armchair" study and into the field. Their approach was participant observation: traveling to a place, living among people and observing their day-to-day lives. The book says Malinowski's Argonauts of the Western Pacific (1922), written after he lived in Trobriand Islands villages, was considered the first modern ethnography.

Business teams borrowed the idea and shrank it. Nielsen Norman Group's guide to field studies (Susan Farrell and Therese Fessenden, January 2024) puts ethnography at the deep end of a range of methods that all happen in the user's natural environment:

The deep end still exists in business research. In the Journal of Business Anthropology (2013), Stephanie Krawinkler describes spending five months over a year and a half as a participant observer in a medium-sized Austrian electrical company for her doctoral research. That is a dissertation, not a product sprint. Most teams reading this should start at the shallow end.

2. Use it when what people say and what they do are likely to differ

Pick ethnographic research when the answer lives in habits, workarounds or the physical setting. NN/g's guide to choosing research methods (Christian Rohrer) frames the split as what people say versus what people do, and notes that self-reported research is limited by what people are aware of and willing to report. It places field studies between the two and recommends leaning toward the behavioral side.

A published example shows the gap. In the same journal issue, anthropologist Maryann McCabe describes a study for a food company with Timothy de Waal Malefyt. Asked about dinner, mothers said they cooked a set repertoire and sometimes tried new dishes. Watched while cooking, they kept adjusting recipes for each person's tastes, such as swapping cinnamon for cumin. That improvising never came up in the sit-down part of the visit, and it turned out to be the insight.

Business teams use this for a few recurring jobs:

• Discovery before design. NN/g says field studies are best done before a design or redesign starts. If you are still choosing a problem, generative research covers where observation fits among the other methods.
• Understanding home and family life around a product. The McCabe study above is one. In another, she describes noticing that participants talked about pets as social connectors in the family, which became a positioning idea for a pet food brand.

The tradeoff is cost. NN/g notes that field studies can be challenging to run, and that field methods "have not become cheaper at the same rate as other usability methods."

3. Pick the lightest version that can see the behavior you care about

Match the method to where the behavior happens and how often. The light versions trade depth for time, and each one misses something.

Contextual inquiry is the site visit done with structure, and that post covers how to run each session.

Home visits are the standard commercial form. Patricia Sunderland and Rita Denny, writing in the Journal of Business Anthropology, say that in US consumer research "an ethnography" has come to mean a three to four hour at-home visit with one recruited person. It usually includes a sit-down interview, a home tour with some demonstration, a shop-along where relevant and collage homework done beforehand. They acknowledge how thin this format can be. It is still a useful baseline, as long as you remember you saw one afternoon.

Diary studies stretch observation over time. NN/g notes they are sometimes called mobile, digital or online ethnography. NN/g's diary study guide (Kim Flaherty, March 2024) suggests matching the length to how often the behavior happens: one week for something done a few times a day, three weeks for a weekly habit. It suggests about 5 to 12 participants for a small discovery project, and it warns that people are more likely to drop out of longer studies. The entries are also only what people choose to log.

Remote observation is cheaper and faster, but NN/g's field studies guide says remote moderated studies are often limited in how much outside context you can see. You will see the spreadsheet. You will not see the sticky note on the monitor bezel that explains it.

4. Run a small one: one question, a week of diaries, then six visits

Here is a made-up example. A startup is building inventory software for independent veterinary clinics. The founders have interviewed practice managers, who all say they reorder supplies "when we run low." Nobody can say what "low" means.

Write one question. "How do clinics notice they need to reorder, and who decides?" If a session drifts into a tour of the X-ray machine, it is drifting.

Recruit six clinics and get permission at each level. NN/g's field studies guide advises getting permission from supervisors, participants and facility managers, and finding an onsite ally to help with logistics. In a clinic that means the owner, the practice manager and whoever runs the front desk. Research participant recruitment covers finding and paying people.

Start with a one-week diary. Ask each practice manager to log every time they notice something is running out and what they did about it. McCabe's food study used the same trick, with participants keeping journals for a week before the home visits. The entries tell you which day to visit and give you something specific to ask about.

Visit for half a day at a busy time. Shadow the practice manager and the front desk, then end with a short interview to check what you saw. Krawinkler describes the same pattern: shadowing a person through a day, then a person-centered interview at the end to clarify observations. In this example, the team might find that Rosa, a practice manager, tracks vaccine stock on a clipboard taped to the fridge, and that the vet techs reorder some items on their own phones. (The clipboard has a pen on a string. The pen is also out of stock.)

Bring one colleague, with a job. McCabe writes that clients who join fieldwork often produce insight, and that they also make the visit harder to manage. NN/g warns that observers can change how participants behave. Give your colleague a notebook and one thing to watch for, and keep them out of the questioning.

Debrief the same day. Write up what you saw before the next visit blurs it. How to take interview notes has a format, and affinity mapping turns six visits of notes into themes.

Short version: one question, about six people, a week of diaries to aim the visits, and a half-day each where they normally do the work.

5. Budget roughly two working weeks of one researcher's time

A small study like the clinic example costs more time than money. Here is a rough plan for that made-up study, for one researcher. The hours are an illustration, not a published benchmark.

That is about two weeks of full-time work, spread over three or four weeks of calendar time because the diary runs first and clinics have their own schedules.

Compare that with five months in the field for one doctoral study. The small version gives up depth: you see one week of diaries and one afternoon per clinic, so seasonal and rare events stay invisible. If the decision depends on those, extend the diary or schedule a second round of visits during the busy season.

6. Interviews are enough when people can recall what you need

Skip the fieldwork when the question is about a decision people remember and can explain: why they switched suppliers, what almost stopped them buying, who signed off. Katarina Graffman, who founded the consultancy Inculture, makes the same point in the Journal of Business Anthropology: fieldwork and participant observation are not always required, and sometimes the job is applying an anthropological eye to facts that already exist.

Interviews also win when you cannot be there. You are unlikely to be invited to a yearly budget meeting, and a rare event will not happen on your visit day. Ask about the most recent one instead, and have the person open the actual documents while you talk. In-depth interviews covers that style.

Choose observation when you hear the same vague phrase in every interview ("when we run low," "it depends") and nobody can unpack it. That phrase usually hides a habit. If your question is about buying in general rather than one workflow, consumer behavior research covers how to combine purchase data, diaries and interviews.

Your next step

Write the one behavior you most need to understand and the place where it happens, then ask three people who do it whether you can watch for half a day next week. Instant Expert can find people matching a description such as "practice managers at independent veterinary clinics" and send them paid invitations for a short call, where you can ask about a visit, and you pay only if someone books. The directory pages for veterinary practice owners, user research experts and consumer insights experts are good places to start.