Diary study: how to run one in UX research

A diary study is a research method where participants log what they do and feel, in the moment, over several days or weeks, so you can see behavior that no single session would catch. Run a diary study when the thing you care about is spread over time (a habit, or an opinion that shifts over weeks) and people would have forgotten the details by the time you interview them. Plan it like a small project: pick when people log, keep it short enough that they finish, pay in installments and read entries every day.

Nielsen Norman Group calls it a qualitative method for collecting behaviors, activities and experiences over time and in context, with participants reporting as things happen. Bolger, Davis and Rafaeli's review in the Annual Review of Psychology puts it this way: people give frequent reports on their daily lives, which capture experience in a way traditional designs cannot.

1. Use a diary study when the behavior happens on its own schedule

An interview asks people to remember. A diary asks them to write it down before they forget. That is the main reason to pick one.

NN/g says single-session methods like interviews and usability tests usually take people out of their own context and cover a short time, so they cannot replace a diary study. It lists the questions diaries answer well:

Habits: when and how often people use something
How attitudes and motivations change over time
Customer journeys that cross several touchpoints
Which device or channel people choose, and why
Outside factors, such as colleagues, competitors and other tools

In psychology research, O'Connor and Ferguson's article for the British Psychological Society lists reducing recall bias among the main aims of diary designs. In plain terms: people's memory of last Tuesday is tidier than last Tuesday was.

Skip the diary when one session would show you everything. If you need to watch a complex task done once, contextual inquiry is faster. If you need someone's reasons for a decision they already made, use in-depth interviews.

2. Pick the logging trigger from how often the behavior happens

The biggest design choice is what tells someone to write an entry. Researchers name three triggers. O'Connor and Ferguson describe them as interval-contingent, event-contingent and signal-contingent; NN/g calls them interval-based, event-based and signal-based.

Trigger	When people log	Use it when	Watch out for
Interval	At set times, such as the end of each day	The behavior is frequent and has no clear start and end	A fixed time skews entries; NN/g notes that reporting each afternoon can miss unusual morning activity
Event	Each time the behavior happens	You need detail on every occurrence and it is rare enough to log each one	People forget to log, especially on busy days
Signal	When you send a prompt, such as a text	Activity varies through the day, or you worry people will forget to report on their own	Prompts arrive at bad moments

You can mix them. O'Connor and Ferguson say the three are not mutually exclusive and suggest considering combinations. A common pairing is event entries plus a short end-of-day check-in that catches what people forgot to log.

Write each entry as three to five short questions about the event that just happened. Ask what they did, where, with what tool, and how it felt. Save the deeper why for the exit interview.

3. Set the length from how long the behavior takes to show up

Run the study for as long as it takes the behavior to happen enough times, and no longer. NN/g ties length to your question, how often the behavior occurs and how long it takes. Its examples: one week for voice assistants used a few times a day, two weeks for buying a phone, three weeks for a meal delivery app used once or twice a week.

If you do not know how often the behavior happens, ask a few people before you set the length. The tradeoff is fatigue: every extra week gives you more data and more people who quietly stop logging.

4. Recruit 5 to 12 people for a focused question, then add extras

NN/g's rough guide to sample size, based on reaching the point where new entries confirm themes instead of changing them:

About 5 to 12 participants for a small discovery project with a similar group and a narrow question
About 12 to 30 for a larger project with a varied group
About 30 to 50 for large or academic studies aiming to generalize

NN/g also recommends overrecruiting because people drop out, which is more likely the longer the study runs. It does not give a percentage, so pick a cushion you can afford, such as two extra people for a study of ten.

Recruit people who do the behavior often. A diary study of someone who orders food delivery twice a year is a very short diary. Research participant recruitment covers channels, and research screener questions helps you confirm frequency before you book anyone.

5. Bookend the logging with a briefing and an exit interview

NN/g recommends a pre-study briefing: explain what to report and how many entries you need, walk through the schedule and incentives, and make sure everyone can use the tool. It also recommends a pilot, which does not need to run as long as the real study, to test your instructions and tool before real participants see them.

Be careful with examples. NN/g warns that sample entries can push participants toward writing only that kind of entry. Show the format, and keep the example dull.

The exit interview is where the diary pays off. NN/g suggests reviewing each person's entries with them, asking follow-up questions and collecting their overall reflections. Read their entries beforehand and pick three or four to ask about ("On day six you skipped the receipt photo. What happened?"). How to take interview notes has a format for the session.

6. Pay more than an interview, in installments, with a minimum number of entries

A diary study asks for more of someone's time than a one-hour call, so pay for all of it. NN/g says diary incentives should be higher than a typical usability session payment, and its broad US guideline is about $40 per hour of total time commitment, adjusted upward for specialized audiences.

NN/g makes three more recommendations:

Set a realistic minimum number of entries, tied to the incentive.
If extra entries earn extra pay, set a maximum too, or you invite made-up entries.
For studies longer than a week, pay in smaller installments at milestones, such as after three days of logging.

Installments help you as much as the participant. Someone who has already been paid once has a reason to keep going. Research incentive budget covers totaling the cost.

7. Read entries every day, because drop-off starts quietly

Logging fatigue is the main way diary studies fail. O'Connor and Ferguson report a criticism from Stone and colleagues that diary methods are often expensive and carry a heavy participant burden. In a 2002 study by Stone and colleagues that they discuss, comparing paper and electronic diaries, the electronic group complied on 94% of occasions and the paper group on 11%. They also cite a counterargument that compliance problems come more from study design and participant motivation than from the format. Either way, the fix is in your design.

Collecting entries as they come in helps. O'Connor and Ferguson say daily returns prevent "hoarding or backfilling," where people fill in missed entries later from memory (which defeats the point of a diary).

Short version: read entries every day and follow up the same day.

NN/g's advice for keeping people engaged:

Tell participants at the start that you will check in, and agree on how.
Send reminders daily or every few days.
Prepare three message templates: thanks for engaged participants, encouragement and help for those falling behind, and a restatement of requirements for those who stop, releasing them if it continues.

Reading as you go also lets you ask about details while they are fresh. A thin entry on Tuesday can become a good one with a single question on Tuesday evening. By Friday, it is gone.

A made-up example: two weeks of receipts

Here is a made-up example. A small company is building expense software for field sales teams. Its interviews keep producing the same tidy answer: "I log expenses at the end of the week." The team suspects the real process is messier, and that receipts get lost long before Friday.

It plans a two-week diary study:

Who: 10 field sales reps who travel at least two days a week, recruiting 12 to allow for drop-off
Logging: an event entry for every work expense (what it was, a photo of the receipt, where the receipt is now) plus a three-minute end-of-day check-in
Minimum: at least one entry on 8 of the 10 workdays
Time asked: a 30-minute briefing, about 15 expense entries at 3 minutes each, 10 check-ins at 3 minutes each and a 45-minute exit interview, which comes to 150 minutes
Pay: NN/g's $40 an hour gives $100 for 2.5 hours. Sales reps are a specialized audience, so the team pays $150 in three parts: $40 after day three, $50 after day ten and $60 after the exit interview
Budget: 12 people at $150 is at most $1,800

By day four, one rep, Dana, has logged five receipts. Her "where is it now" answer for three of them is the sun visor of her car. That is the detail an interview would have missed: the problem is the gap between paying and photographing, which a weekly reminder will never fix. The exit interview asks Dana what happens to the visor receipts on rainy days.

Your next step

Write down the one behavior you want to see, how often it happens and the trigger that fits it. Then draft five entry questions and pilot them with two people for three days before the real study.

If the people you need are outside your customer list, Instant Expert can find people matching a description such as "field sales reps who travel at least two days a week" and send them paid invitations for a short call, such as a screening call before the study, and you pay only if someone books. The directory pages for user research experts and UX consultants can help with study design.