Skip to main content

UX and research · Intermediate · 17 min

Run a two-week diary study

By Padding · Updated 18 Nov 2026
Any online form tool, a spreadsheet and a messaging app participants already use · November 2026

You will have run a two-week diary study on how people try to see a doctor, with a tested entry form, a reminder schedule, a mid-study check-in and a tidy sheet of entries ready to analyse.

On this page

An interview asks people to remember. A diary study asks them to write it down while it is still happening. That difference matters when the thing you care about is spread over days: a symptom that gets worse, a phone line that is engaged at 8am, a booking page that times out, a decision to give up and go to the pharmacy instead. By the time you interview someone a fortnight later, those moments have collapsed into "it was a hassle".

The cost is that a diary study runs for weeks, and you are not in the room. Participants forget, get bored, write one-word entries, or stop altogether around day five. Most of the work in this tutorial is about that: making entries quick to give, prompting at the right moments, and noticing drop-off early enough to do something about it.

The running example is a study for a team designing a way for patients to book GP and clinic appointments. The question: "What happens, from first deciding to see a doctor to actually being seen, and where does it stall?" Participants log every attempt to see a doctor over 14 days. Swap in your own question as you go; the structure holds for anything people do repeatedly.

By the end you will have a one-page study plan, an entry form you have tested, a reminder schedule, a tracking sheet you check daily, a mid-study check-in, and a cleaned dataset of entries.

Step 1 · Decide what counts as an entry

A diary study lives or dies on its trigger: the event that tells a participant "write something now". If the trigger is vague ("anything about your health"), you get vague entries, or none. If it is too narrow, nothing happens for days and people lose the habit.

Open a document called "Diary study plan" and write three things at the top.

  • The question, in one sentence.
  • The trigger, as a behaviour someone can recognise in the moment. For the example: "Any time you do something to try to see a doctor or nurse: phone, visit a website or app, walk in, ask someone to book for you, or decide not to." Including "decide not to" matters; the attempts people abandon are often the most useful.
  • The window: 14 days, with a named start and end date. Start on a weekday so the first entries come while your briefing is fresh.

Then decide what happens on days with no trigger. In a health study, many participants will have quiet days. Ask for a short "nothing today" check every two or three days, so silence means "nothing happened" rather than "I forgot".

Write down the minimum you need from each participant to keep their data: for example, at least six entries or check-ins across the two weeks, including at least one in the second week. Decide this now, before you see the data, so you do not bend it later.

Check it worked

Your plan has one question, one trigger that names concrete actions (including giving up), a 14-day window with dates, a rule for quiet days and a written minimum for a usable participant. Read the trigger to someone outside the project and ask them to name two things that would count; if they hesitate, rewrite it.

Step 2 · Build an entry form that takes under two minutes

People will fill in the form standing in a queue, on hold, or in bed with a temperature. Every extra question costs you entries. Aim for five to seven questions, most of them taps rather than typing.

Create a new form in your form tool, set it to collect responses into a spreadsheet, and add these questions in this order:

  1. Your participant code (short text, for example P04). Codes rather than names keep the sheet private from the start.
  2. When did this happen? Date and rough time. Entries are often logged later than the event, so ask; do not rely on the submission timestamp.
  3. What did you do? Single choice: phoned the surgery or clinic, used a website or app, walked in, asked someone else to book, used a pharmacy instead, decided not to do anything, other.
  4. What happened? Single choice: got an appointment, was told to try again later, was put on a waiting list, was redirected somewhere else, gave up, still waiting.
  5. How did it go? A five-point scale from "very hard" to "very easy".
  6. Tell us about it. Long text, optional, with the hint: "What happened, what you thought, what you did next. A voice note or a photo is fine too."
  7. Anything you want to add? Optional.

Keep the open question optional. Making it required produces "ok" and "fine". People who have something to say will say it; the multiple-choice answers still give you a usable record when they do not.

Add a second, shorter form (or a branch at the start of the same form) for the quiet-day check: participant code, date, and "Did you do anything to try to see a doctor today? Yes, I have logged it / No / I forgot to log something". That last option recovers missed entries without guilt.

Check it worked

Fill in the form three times on your phone as if you were a participant: once for a successful booking, once for giving up, once as a quiet day. Time each one. All three take under two minutes, and the responses appear as separate rows in the linked spreadsheet with the participant code in its own column.

Step 3 · Pilot the form with one person for two days

Before you brief anyone, give the form to one person who matches your participants but is not in the study: a colleague's relative, or a spare from recruitment. Ask them to use it for two days and log at least two real or recalled attempts.

Then talk to them for ten minutes. Ask which question made them pause, which option was missing, and whether they would keep doing this for two weeks. Typical fixes after a pilot:

  • An option people keep choosing "other" for (in the example, "used the NHS app" or "messaged the surgery" might need their own option, depending on where you run the study).
  • A scale that runs the wrong way for how people think.
  • A question that duplicates another.

Change the form now. Changing questions in the middle of the study splits your data into two incompatible halves.

Check it worked

You have a short note in your plan headed "Pilot changes" listing what you changed and why, and no response option in the pilot data was chosen as "other" more than once without a fix.

Step 4 · Plan prompts and reminders

Reminders are not nagging if they are predictable and short. Write the schedule in your plan before you start, so you send the same messages to everyone.

Use the channel participants already check (a text message, or a messaging app they named at recruitment), not email, which is easy to ignore for two weeks.

A schedule that works for a 14-day health study:

  • Day 1 morning: welcome, the form link, the trigger in one line, and your contact.
  • Every two days, early evening: a short nudge with the quiet-day link: "Did you try to see a doctor or nurse today? If so, log it here. If not, tap here to tell us."
  • Day 4: a personal note to anyone with no entries yet (see the tracking step).
  • Day 7: the mid-study check-in.
  • Day 10: a nudge that varies the prompt: "Think back to the last time you needed a doctor this week. What nearly stopped you?"
  • Day 13: "Two days left", and a reminder of the thank-you.
  • Day 14 evening: close, thank them, and say when the thank-you will arrive.

Varying the prompt in week two is deliberate. Entries shrink over time as the novelty wears off; a fresh question gives people something new to say.

Write every message out in full in your plan, with the day and time it goes. If your tool lets you schedule messages, schedule them; if not, set calendar reminders for yourself.

Check it worked

Your plan contains a table of every message: day, time, channel, exact text. No two consecutive days have more than one message, and the week-two prompts are different from the week-one prompts.

Step 5 · Brief participants with a practice entry

Run a 15-minute call (or a short in-person meeting) with each participant, or with small groups, the day before the study starts. A written welcome alone is not enough; most drop-off in the first three days comes from people who were not sure what counted.

Cover four things:

  1. What the study is for, in one sentence, and that there are no right answers.
  2. The trigger, with two examples, including giving up.
  3. The form: ask them to open it on their phone now and log a practice entry about the last time they tried to see a doctor. Watch or listen while they do it.
  4. What you need from them (your minimum from the first step), how they will be thanked, and that they can stop at any time.

Confirm consent here if you have not already, including how long you will keep their entries and who will see them. Give each participant their code and ask them to save the form link to their home screen.

Delete the practice entries from the sheet afterwards, or mark them in a "practice" column, so they do not end up in the analysis.

Check it worked

Every participant has a practice entry in the spreadsheet under their own code, marked as practice. Anyone who could not submit one has a follow-up noted before day 1.

Step 6 · Track entries daily and act on drop-off

Add a second tab to your response spreadsheet called "Tracker". List participant codes down the side and the 14 study days across the top. Each morning, fill in each cell from yesterday's responses: E for a logged entry, Q for a quiet-day check, blank for nothing. Add a column for total entries and one for "last heard from".

This takes five minutes a day and is the main thing that keeps a diary study alive. Watch for three patterns:

  • Silent from the start. No entries by day 3. Send a personal message, not the group nudge: "Hi, just checking the form is working for you. Has anything come up yet?" Often the link broke or they were unsure what counted.
  • Fading. Entries on days 1 to 4, then nothing. Mention one thing they wrote that was useful and ask if anything has changed. People keep going when they know someone reads their entries.
  • Thinning. Entries keep coming but the open answers shrink to a few words. That is your cue for the varied prompt on day 10, or a short follow-up question to that person about a specific entry.

Read the entries each day too, not only count them. When an entry is unclear ("tried again, same thing"), reply within a day with one question. Waiting until the debrief means they will not remember.

Do not chase more than twice in a row. If someone stays silent after two personal messages, note it and let them be; they may come back, and pressure costs you the people who are still active.

Check it worked

Your Tracker tab has a row per participant and a filled-in column for every day so far, a "last heard from" date for each person, and a note beside anyone you messaged personally saying what you sent and when.

Step 7 · Run the mid-study check-in

On day 7, run a 10 to 15 minute call with each active participant. This is not an interview about the topic; it is a check that the study is capturing what is happening.

Before the call, reread that person's entries and pick two to ask about. Then ask:

  • "Walk me through this one" for one entry with a short open answer. You will often learn the important part was not written down.
  • "Has anything happened that you did not log? Why not?" This tells you whether the trigger is missing a kind of attempt.
  • "Is anything about the form annoying?" Fix wording problems only if the fix does not change what a question means.
  • "Are you OK to keep going for another week?"

Take notes in a third tab called "Check-ins", one row per participant, with the date and what you learned. If several people say they did not log the same kind of event (for example, asking a pharmacist for advice), send everyone a one-line clarification of the trigger that evening, and record the date of the change in your plan.

Check it worked

Your Check-ins tab has a row for every participant still active on day 7, each with at least one entry they expanded on. Your plan records any trigger clarification with the date it was sent.

Step 8 · Close the study and tidy the dataset

On day 14, send the closing message, then close the form so late entries do not arrive after you start analysis. Send thank-yous within the time you promised.

Then make a clean copy of the response sheet, called "Diary dataset", and tidy it:

  1. Remove practice entries and any test entries you made.
  2. Add a column "event date" from the participant's own answer, and sort by participant then event date. Submission time is not when things happened.
  3. Add a column "usable participant" (yes or no) using the minimum you wrote in the first step. Do not change the rule now.
  4. Add a column "expanded in check-in" and paste in anything participants added on calls, so the entry and its explanation sit together.
  5. Remove anything identifying from open answers: names of doctors, surgeries, addresses, phone numbers. Replace them with a label such as [surgery].
  6. Count entries per participant per week. A big drop in week two is normal; write the numbers down so you can say how complete the data is when you report.

Keep the raw sheet private and unchanged, and analyse from the copy.

Check it worked

Your "Diary dataset" sheet has one row per entry, sorted by participant and event date, with no practice rows, no real names or places in open answers, a usable-participant flag, and a small summary of entries per participant per week. The original response sheet is untouched.

Next steps

Your entries are ready to sort into patterns. Pull the open answers and check-in notes onto sticky notes and group them in Build an affinity map from interview notes; the same method works for diary entries, with the multiple-choice columns giving you counts to set beside each theme.

If the diary shows where attempts stall but not why, follow up with a few participants in their own setting using Run a contextual inquiry.

For your next study, start from Write a one-page research plan, and use Recruit participants for user research to over-recruit by two or three people, since some diary participants will drop out whatever you do.

Need this done for you?

Padding does this work for growing companies, from the first diagnosis to a system that keeps working.

Digital Transformation

Get new tutorials

New tutorials, about twice a month.