Skip to main content

UX and research · Beginner · 13 min

Write a screener survey to find the right participants

By Padding · Updated 18 Nov 2026
Any form builder and a spreadsheet · November 2026

You will have a screener of eight questions that finds people who booked a clinic visit in the last six months without hinting at the answer you want, plus written rules for who qualifies, who is turned away and who fills a quota.

On this page

A screener is the short questionnaire people fill in before you invite them to an interview or a usability test. Its only job is to sort applicants into "invite" and "do not invite". Get it wrong and you spend five sessions listening to people who have never done the thing you are studying, or to people who have learned how to pass screeners.

The second group is the bigger risk. Some people take part in paid research often and know what screeners look for. If your questions make the right answer obvious ("Have you booked a clinic appointment online recently? Yes / No"), they will pick it, and you will not be able to tell them apart from the people you wanted.

This tutorial uses Larkfield Clinic, a fictional community clinic that is redesigning its appointment booking app. The team wants to interview people who booked a clinic visit in the last six months, by any method. You will write the profile first, then the questions, then the rules, and finally try the screener on a few people before it goes out. Swap in your own study as you go.

Step 1 · Write the profile before any question

Open a spreadsheet and make a tab called Profile. Write three lists, one per column:

  • Must have. Behaviour the person has to show, with a time frame. For Larkfield: "Booked an appointment at a clinic, GP surgery or health centre in the last six months, for themselves or someone they care for."
  • Mix. Things you want a spread of, not a single answer. For Larkfield: how they booked (phone, website, app, in person), age band, and whether they booked for themselves or for someone else.
  • Exclude. People whose answers would mislead you. For Larkfield: people who work in user research, design, market research or software; people who work in clinic administration or reception (they know the booking system from the inside); and people who have taken part in several paid studies recently.

Keep the must-have list short. Every extra must-have shrinks the pool and slows recruiting. If a criterion would be nice but would not change what you learn, move it to Mix or delete it.

Write the criteria as behaviour, not attitudes or labels. "Booked a visit in the last six months" can be checked against a memory. "Is comfortable with technology" or "cares about their health" cannot, and anyone can claim them.

Check it worked

Your Profile tab has three columns. The Must have column has no more than three rows, and every row names something the person did, with a time frame. No row in any column uses words like "comfortable", "interested in" or "values".

Step 2 · Turn each must-have into a question with several honest answers

A must-have becomes a question where the qualifying answer is one choice among several plausible ones. Yes/no questions are the weak point of most screeners: a person who wants the reward can see which answer gets them in.

Compare these two versions of the same must-have:

  • Weak: "Have you booked a medical appointment in the last six months?" Yes / No.
  • Stronger: "When did you last book an appointment at a clinic, GP surgery or health centre?" In the last month / 1 to 6 months ago / 6 to 12 months ago / More than a year ago / I have never booked one myself / I don't remember.

The stronger version has no answer that looks like the "good" one. Every option is a normal thing for a person to say, so there is nothing to aim for. It also gives you more than you asked for: you can see who booked recently enough to remember the details.

Do the same for anything in Mix that needs a question. For "how they booked", list the methods as equal options rather than asking "Did you use an app?":

  • "How did you make that booking?" By phone / On a website / In an app / In person at the clinic / Someone else booked it for me / Other (please say).

Check it worked

Read each question in your draft. None of the must-have questions can be answered yes or no, each has at least four answer options, and each offers a "don't remember" or "other" option so nobody is pushed into a wrong answer.

Step 3 · Hide the topic and the answer you want

If the screener's title or first question says "clinic booking app", applicants know what you want before they read a single option. Hide the topic in two ways.

Describe the study broadly. In the form's introduction, say what taking part involves and what people receive, but name the subject in general terms: "We are talking to people about how they arrange everyday appointments and services." That is true, and it does not tell anyone to say "clinic".

Put the qualifying activity in a list of similar ones. Open with a multiple-choice question where the clinic booking is one item among several:

  • "Which of these have you done in the last 12 months? Tick all that apply." Booked a haircut or beauty appointment / Booked an appointment at a clinic, GP surgery or health centre / Booked a table at a restaurant / Booked a car service or repair / Renewed an ID document or licence / None of these.

Someone who ticks every box is a warning sign. Real people rarely do all of them in a year, and a person trying to qualify for anything often does. Note it for the rules step.

Then ask the "when did you last" question from the previous step only after this list. By then the applicant has seen five kinds of appointment and does not know which one matters.

Check it worked

Show your introduction and the first two questions to a colleague who does not know the study. Ask them: "What do you think we are researching, and which answers do you think get someone invited?" If they name clinic booking from the introduction alone, or pick out the qualifying answers straight away, rewrite until they cannot.

Step 4 · Add questions that screen out professional testers and insiders

Two short questions catch most of the people you listed under Exclude. Ask them without saying why.

Work and household. "Do you, or does anyone in your household, work in any of these areas?" Market research / User experience or design / Software development / Clinic or hospital administration or reception / Journalism or media / None of these. Any answer other than "None of these" is a reject, and the list gives no clue which jobs are the problem.

Recent research. "In the last six months, how many times have you taken part in research for a reward, such as interviews, product tests or focus groups?" None / Once / Twice / Three times or more. Use bands, not a yes/no, so "None" does not stand out as the answer you want.

Then add one open question. It is your best check for both honesty and fit:

  • "Please describe the last time you booked a clinic or GP appointment: what it was for (you do not need to give medical details), how you booked it and anything that was difficult."

People who really booked recently give specifics: a time of day, a phone queue, a form that would not load. People inventing an answer give one vague line. The open answer also shows you who can describe an experience in their own words, which matters for an interview. Say "you do not need to give medical details" in the question, because this is health-adjacent research and you have no reason to collect diagnoses.

Check it worked

Your draft has a work-and-household question with "None of these" as the only accepting answer, a past-research question with at least four bands, and one open question asking for a specific recent experience. The open question says that medical details are not needed.

Step 5 · Add the mix and logistics questions, then cut to eight

Finish with what you need to balance the group and book the session:

  • Age band. Under 25 / 25 to 39 / 40 to 59 / 60 or over / Prefer not to say. Use bands, not birth dates.
  • Availability. "Which of these times could you join a 45-minute video call in the week of [dates]?" Weekday mornings / Weekday afternoons / Weekday evenings / Saturday / None of these.

Now put the screener in this order and count the questions:

  1. Which of these have you done in the last 12 months? (multiple choice, includes the clinic option)
  2. When did you last book an appointment at a clinic, GP surgery or health centre?
  3. How did you make that booking?
  4. Was the appointment for you, or for someone you care for?
  5. Do you, or does anyone in your household, work in any of these areas?
  6. In the last six months, how many times have you taken part in research for a reward?
  7. Please describe the last time you booked a clinic or GP appointment. (open)
  8. Which age band are you in?

Availability can go on a separate booking form sent only to people who qualify. That keeps the screener at eight questions and avoids asking everyone for details you will only use for a few.

The order matters. The questions most likely to reject someone (1, 2, 5, 6) come early, so applicants who do not fit spend less time. The open question comes late, once the applicant has already given you the facts you will check it against.

If you have more than eight questions, cut. For each one, ask: "Would a different answer change whether I invite this person, or which quota they fill?" If not, delete it. Questions you are curious about belong in the interview, not the screener.

Check it worked

Your form has six to eight questions in the order above, or in an order where rejecting questions come before the open question. Every remaining question either decides invite or reject, or places the person in a mix group. You can name which for each one.

Step 6 · Write the qualification rules

Add a tab called Rules to your spreadsheet. List every question with every answer option in its own row, and label each option with one of four outcomes:

  • Accept: this answer keeps the person in.
  • Reject: this answer removes the person.
  • Mix: this answer places the person in a group you want a spread of.
  • Flag: this answer needs a human to look before deciding.

For Larkfield the key rows look like this:

  • Q1: clinic option ticked, Accept. Clinic option not ticked, Reject. Every option ticked except "None of these", Flag.
  • Q2: In the last month or 1 to 6 months ago, Accept. Any other answer, Reject.
  • Q3: each method, Mix. "Someone else booked it for me", Reject, because they did not do the booking you are studying.
  • Q4: for me / for someone I care for, Mix.
  • Q5: None of these, Accept. Any other option, Reject.
  • Q6: None or Once, Accept. Twice or Three times or more, Reject.
  • Q7: specific answer that matches Q2 and Q3, Accept. Vague, copied or contradicting Q2 or Q3 (says "last week on the app" after choosing "6 to 12 months ago" and "by phone"), Reject.
  • Q8: each band, Mix. Prefer not to say, Accept, never Reject.

Below the list, write the quotas. For six interviews: at least two people who booked by phone and at least two who booked online (website or app), at least one who booked for someone they care for, and no more than three from any one age band. Quotas stop the easiest people to reach from filling every slot.

Last, write the tie-break rule: when more people qualify than you have slots, pick by quota first, then by the most specific answer to Q7. Writing this down before responses arrive stops you choosing people because their answers sounded pleasant.

Check it worked

Every answer option in your form appears in the Rules tab with exactly one of Accept, Reject, Mix or Flag. Your quotas add up to no more than the number of sessions you plan to run, and the tie-break rule is written under them.

Step 7 · Pilot the screener with three people

Before sending it out, ask three people to fill it in and read their answers against your rules. Try to include one person who should qualify and one who should not.

Look for three problems:

  • A wrong result. Someone who fits the profile is rejected, or the reverse. Usually an answer option is labelled wrongly or is missing ("I booked by email" with no Email or Other option).
  • Confusion. A pilot participant asks what a question means, or picks "Other" and writes something one of your options already covers. Reword the question or the option.
  • A visible target. Ask each person afterwards which answers they thought would get them picked. If they guess correctly, the topic or the answer is still showing. Go back to the step on hiding the topic.

Once the screener goes live, check the first ten responses against the Rules tab by hand before you trust it. A copy-paste mistake in an answer label can quietly reject everyone.

Check it worked

You have three pilot responses in your spreadsheet, each marked Accept or Reject using only the Rules tab, and each outcome matches what you expected for that person. You have a short list of the changes you made after the pilot, even if the list says "none".

Next steps

You now have a screener and the rules to read it. Next, put it to work:

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.