Some questions can be answered in a one hour session: whether a label makes sense, where a flow breaks, how someone reacts to a prototype. But a lot of the behaviour that matters unfolds over days and weeks, not minutes. Habits form, novelty wears off, workarounds harden into routine. A diary study is the method built for exactly this: instead of asking people to remember what they did, you ask them to record it as it happens, in their own lives, over time.
Run well, a diary study gives you something no interview can: behaviour captured in the moment, before memory has tidied it up. Run badly, it gives you dropouts, a pile of one word entries, and an analysis backlog you resent. This guide is about landing on the right side of that line.
When a diary study beats a snapshot
Interviews and usability tests are snapshots. They are excellent at depth, but they rely on either an artificial moment (the session) or a reconstructed one (the participant's memory). And memory is a poor witness: people generalise, they remember peaks and endings rather than the ordinary middle, and they quietly smooth their past behaviour into something tidier than it was.
Longitudinal questions need a longitudinal method. Reach for a diary study when your question sounds like:
- How does use change between day one and week three?
- What actually triggers people to open the app, or abandon the task?
- How does this fit into the rhythm of someone's day, alongside everything else competing for attention?
- What are people feeling at the moment of use, not in retrospect?
The method sits in the same family as ethnographic research: you are studying people in their own context rather than yours. The difference is that participants do the observing on your behalf, asynchronously, which is what makes weeks-long timescales affordable.
Structured or open entries
The core design decision is what an entry looks like.
Structured entries give participants fixed fields: what were you doing, where were you, how frustrating was it on a scale of one to five. They are quick to complete, comparable across people and across days, and far easier to analyse. The cost is that they can only capture what you thought to ask about.
Open entries ask something looser: "tell us about a moment today when the tool got in your way." They surface the unexpected, in the participant's own words, and they tend to produce the material that carries the final report. The cost is effort, variable quality, and a heavier analysis load.
Most good diary studies are a hybrid: two or three fixed fields for the backbone, plus one open question for the surprises. And let people answer in whatever medium is easiest in the moment · a photo, a screenshot, a thirty second voice note. Lowering the effort of capture raises both the quantity and the honesty of what you get back.
Prompts and cadence
When people log matters as much as what they log. There are three broad patterns:
- Interval based: entries at a fixed time, usually the end of the day. Good for reflection and routine behaviour; weakest on in-the-moment detail.
- Signal based: you ping participants at random or semi-random times and ask what is happening right now. This is the experience sampling tradition, and it is the least distorted by memory.
- Event based: participants log whenever the thing you care about happens. Best for behaviour with its own natural trigger, like every time you pay for something, or every time you give up on a search.
Match the cadence to the rhythm of the behaviour, then keep each entry tiny. A two minute entry twice a day survives a fortnight; a ten minute form does not. On duration, one to two weeks suits most product questions: long enough for a pattern to show, short enough for compliance to hold.
Whatever you choose, pilot the prompts on yourself or a colleague for two days first. Every ambiguous word in a prompt costs you a week of muddy data.
Keeping participants engaged
Dropout is the characteristic failure of a diary study, and most of it is preventable.
- Start with a proper onboarding conversation, not just an instructions email. Walk through one example entry together, so expectations about quality are set on day one.
- Respond like a human. A short personal reply to an early entry does more for completion than any incentive tweak. Silence teaches people that nobody is reading.
- Watch the first 48 hours closely. Confusion shows up immediately; fix it immediately.
- Structure incentives across the study, with a meaningful completion bonus, rather than one flat payment at the end.
- Over-recruit slightly. Some attrition happens even in well run studies, so plan your numbers with it in mind, the same way you would think about sample size in any qualitative work.
The analysis workload, honestly
Here is the part method guides tend to skim past: ten participants making two entries a day for two weeks is 280 entries, plus photos and voice notes. If you first open the data after fieldwork ends, you will drown.
Analyse on a rolling basis instead. Read entries as they arrive, tag them lightly, and keep a running note of emerging patterns. The structured fields give you quick counts; the open entries need proper qualitative coding, so budget real time for it, ideally as much time as the fieldwork itself.
Then close with debrief interviews. This is where a diary study quietly doubles its value: instead of a cold interview that leans on recall, you sit down with two weeks of the participant's own moments and say "tell me more about Tuesday." The diary provides the specifics; the conversation provides the meaning.
The pattern running through all of this is restraint. Small entries, few prompts, a modest window of time, analysis that starts on day one. Diary studies punish ambition and reward discipline, and the researchers who get consistent value from them design for the participant's fortnight, not for the team's wish list.
People of Research is a community and newsletter for UX researchers, ethnographers, and anyone who studies what people do in their own lives, not just in sessions. Join us and get the good stuff in your inbox.