Retour

Screener Questions: Examples, Logic and Trap Questions

Matthieu Saussaye

A survey is only as good as the people who answer it, and screener questions are the only door you control.

Screener questions are the short block at the start of a survey that decides who qualifies, who is sent away and who fills which quota. Get them wrong and every chart after them describes the wrong population. Get them right and a 300-person study can carry a real decision. With SmartInterview, the screener is drafted with the rest of the questionnaire from your objective, and screen-outs and quotas are enforced live while the study runs, which makes it the fastest way to run a properly screened study.

The Short Answer

  • Purpose: screener questions filter respondents against your target definition (who buys, uses, decides or lives where) before they see the main questionnaire.

  • Two kinds of exit: a terminate (screen-out) removes people who are not your target; a quota full removes people who are in the target but belong to a group you already have enough of.

  • Do not reveal the target. Ask neutral, multi-option questions ("Which of these have you bought?") instead of yes/no questions that tell people the right answer.

  • Add one trap. A red herring option (a brand or product that does not exist) and a plausibility check catch most people who claim everything to qualify.

  • Keep it short: 3 to 7 questions, about a minute. Each extra criterion lowers incidence and raises cost.

  • Do the math first: required starts = target completes divided by incidence rate. 300 completes at 20% incidence means about 1,500 people entering the screener.

Screener Question Types and When to Use Them

Screener type

Example question

Typical logic

Watch out for

Category usage

Which of these products have you bought in the past 3 months?

Terminate if target category not selected

Yes/no versions invite people to say yes

Brand usage

Which of these brands have you used in the past 12 months?

Route users vs non-users into quota cells

Include a fake brand as a red herring

Demographics

What is your age? / In which canton do you live?

Quotas on age, gender, region

Ask age as a number or bands, not "Are you 25-44?"

Role / decision power (B2B)

What is your role in choosing software for your team?

Terminate "no involvement"

Job titles alone overstate authority

Company profile (B2B)

How many people work in your organisation?

Quota by company size

Self-reported revenue is unreliable

Industry exclusion

Do you or anyone in your household work in any of these industries?

Terminate market research, advertising, the client's sector

Hide the exclusion among neutral options

Past participation

Have you taken part in a study about this topic in the past 6 months?

Terminate recent participants

Matters most in qualitative work

Attention / trap

Have you ever been licensed to operate a nuclear submarine?

Flag or terminate implausible answers

Use one or two, not five

Most screeners are single-select or multiple-choice questions. Avoid yes/no questions for qualification: they are fine in the main survey, but in a screener they show the respondent exactly which answer gets them in.

How Screening Works: Definitions, Logic and the Math

What a screener question is

A screener question is a question whose answer decides whether a respondent continues. It is not there to learn anything about the market; that is the job of the main questionnaire. Everything in the screener should map to one line of your target definition. If you cannot say which criterion a question tests, delete it. Screening applies the same way whether you recruit from a research panel or invite your own customers, although customer lists usually need fewer criteria because you already know who they are.

Terminate vs quota full

These two exits look identical to the respondent and are completely different for your data. A terminate (also called screen-out or disqualify) means the person is outside the target: a non-user in a user study, a student in a study of IT buyers. A quota full (overquota) means the person is inside the target, but their cell is already complete: you wanted 150 women and already have 150. Panels usually report and often pay these statuses differently, and you should track them separately: a high terminate rate tells you the incidence is lower than you assumed; a high quota-full rate tells you the sample is arriving unbalanced.

Quotas protect the shape of the sample. They do not make a non-probability sample representative by themselves; see sampling methods for the difference between quota and probability samples.

Incidence rate: the number that sets your cost

Incidence rate (IR) is the share of people entering the screener who qualify. It drives both cost and fieldwork time, which is why panels quote by incidence.

Required starts = target completes / incidence rate.

Study

Target completes

Incidence

Required starts

Coffee drinkers, Switzerland

300

70%

about 430

Bought plant-based protein, past 3 months

300

20%

1,500

IT decision makers, 50-249 employees

150

8%

about 1,875

Owners of a specific e-bike model

100

2%

5,000

Then add the people you will remove for quality. If a trap question and speed checks remove another 10% of qualified respondents, the plant-based study needs 300 / (0.20 x 0.90), about 1,667 starts. Each criterion multiplies: if 40% of a population drinks coffee and 50% of those buy pods, incidence is 0.40 x 0.50 = 20%. Adding "and bought our brand" at 30% drops it to 6%, and the required starts for 300 completes jump from 1,500 to 5,000.

Why traps are now necessary

Pew Research Center tested this directly in a 2022 experiment: in an online opt-in sample, 12% of adults under 30 said they were licensed to operate a class SSGN nuclear submarine, an answer whose true rate rounds to zero (Pew Research Center, March 2024). Pew describes these as "bogus respondents" who answer to finish quickly and collect a reward. If people will claim a submarine licence, they will certainly claim to have bought your category. That is a bias problem you fix at the door, not in the analysis.

Ready-to-Use Screener Questions, Step by Step

Below is a full screener in the order you would field it, with B2C and B2B variants. Copy the wording, change the categories, keep the structure.

1. Country and region (B2C and B2B)

In which country do you currently live?

  • Switzerland

  • Germany

  • France

  • Austria

  • Italy

  • Other country

Logic: terminate "Other country"; follow with canton or region for regional quotas.

2. Age (B2C)

What is your age?

  • Under 18

  • 18-24

  • 25-34

  • 35-44

  • 45-54

  • 55-64

  • 65 or older

Logic: terminate under 18; quota on the bands you need. Never ask "Are you between 25 and 44?"

3. Industry exclusion (B2C and B2B)

Do you, or does anyone in your household, work in any of the following? Select all that apply.

  • Market research or advertising

  • Journalism or media

  • Food and beverage manufacturing

  • Banking or insurance

  • Education

  • Healthcare

  • None of these

Logic: terminate the first option and the client's own sector (here, food manufacturing). The other options are neutral cover so the excluded sector is not obvious.

4. Category purchase with a red herring (B2C)

Which of the following have you bought for yourself in the past 3 months? Select all that apply.

  • Protein bars

  • Plant-based milk

  • Vitamin supplements

  • Energy drinks

  • Frozen ready meals

  • Velorin sports drops

  • None of these

Logic: qualify on "Vitamin supplements"; "Velorin sports drops" does not exist, so anyone who selects it is flagged or removed. Make sure your fake name is not a real product in any target market before you field it.

5. Purchase frequency (B2C)

How often do you buy vitamin supplements?

  • At least once a month

  • Every 2-3 months

  • 2-3 times a year

  • Once a year or less

Logic: quota "heavy" (monthly) vs "light" buyers, or terminate the last option if you only want regular buyers.

6. Decision role (B2B)

What best describes your role when your organisation chooses a new software tool for your department?

  • I make the final decision

  • I recommend and influence the decision

  • I am consulted but do not influence it much

  • I am not involved

Logic: qualify the first two; terminate the last two. This works better than filtering on job title.

7. Company size (B2B)

Approximately how many people work in your organisation, across all locations?

  • 1-9

  • 10-49

  • 50-249

  • 250-999

  • 1,000 or more

  • I don't know

Logic: quota by size band; terminate "I don't know" for decision-maker studies, since a real buyer knows roughly.

8. Plausibility check (B2C and B2B)

Which of the following have you done in the past 12 months? Select all that apply.

  • Travelled by train

  • Bought something online

  • Run a full marathon

  • Visited Antarctica

  • Piloted a commercial aircraft

  • None of these

Logic: anyone selecting both of the last two is almost certainly not answering honestly. Flag rather than terminate on a single implausible choice.

9. How to do it in SmartInterview

In SmartInterview, you describe the objective ("understand why women 30-55 in Switzerland who buy supplements choose one brand over another") and the AI drafts the full questionnaire, screener included, with a quality score that flags issues before launch. You set terminate and quota rules, and screen-outs and quotas are enforced live, so you do not pay for people outside your target or for an overfilled cell. You can mix your own customer list (a reusable link, every response tagged by source) with an audience recruited in Switzerland and Europe. Once people qualify, the AI asks a follow-up when an open answer is vague, respondents can answer by text or voice, and open answers are automatically coded into themes linked to the verbatims, with sentiment and cross-tabs by any screener variable. Results typically arrive in about 48 hours.

Where I draw the line

I keep screeners strict on who and loose on everything else. Strict on the criteria that define the decision: if the question is about supplement buyers, a non-buyer's opinion is noise, however articulate. Loose on the rest: I do not add "and uses a smartphone app" or "and shops at Migros" unless the decision really depends on it, because each of those criteria quietly divides the incidence and multiplies the cost.

I also refuse to use the screener as a mini-survey. If a question does not decide whether the person continues, it goes after qualification, where the answer is honest and the person is already engaged. And I always run one trap. Not five: too many tricks annoy honest respondents and teach professional ones what to avoid.

Matthieu Saussaye, Founder of SmartInterview.

Screener Checklist Before You Launch

  1. Write the target definition in one sentence, then map every screener question to a part of it.

  2. Estimate incidence for each criterion and multiply. Calculate the required starts and check the budget.

  3. Order from broad to narrow: country, age, industry exclusion, category, brand, then role or frequency. People leave early, which saves cost.

  4. Replace every qualifying yes/no with a list of options plus "None of these".

  5. Randomise option order on lists of brands and categories (keep "None of these" last).

  6. Add one red herring and check that the fake name does not exist in any target market.

  7. Separate terminate and quota-full statuses and decide what each exit page says.

  8. Test the logic with five dummy profiles: one that qualifies, one per terminate rule, one that hits a full quota.

  9. Soft launch to 10% of the sample and compare the real incidence with your estimate before releasing the rest.

Common mistakes

  • Leading screeners. "Do you buy organic food regularly?" gets more yes answers than a neutral list where organic is one option among several.

  • Revealing the topic too early. An intro that says "a study for supplement buyers" tells everyone what to claim. Describe the topic in general terms.

  • Too many criteria. Every extra filter multiplies incidence down; four modest criteria can leave you below 5%.

  • Treating quota full as a terminate. You lose the information that the target was there and the sample was simply unbalanced.

  • Screening on attitudes. Filtering on "very satisfied" customers before measuring satisfaction makes the satisfaction score meaningless.

  • No back button control. If respondents can go back after a terminate, some will change their answers until they get in.

  • Forgetting the length. A three-minute screener raises drop-out before the survey even starts, one of the reasons response rates are falling.

A properly screened sample is what made the Pharmalp study usable: the Swiss natural health brand validated pricing and positioning of four new supplement concepts with people who actually buy in the category. For the rest of the questionnaire, see concept testing methods and the survey questions guide.

Frequently Asked Questions

What are screener questions in a survey?

Screener questions are the first questions of a survey that check whether a respondent matches the target audience. Depending on the answers, the person continues, is screened out, or is stopped because their quota group is already full.

How many screener questions should a survey have?

Usually 3 to 7, taking about a minute. Each question should test one part of your target definition. More questions raise drop-out and, if they add criteria, lower the incidence rate and raise the cost.

What is the difference between a screen-out and a quota full?

A screen-out (terminate) removes someone who is not in the target. A quota full removes someone who is in the target but belongs to a group that already has enough completes. Track them separately: they tell you different things about your incidence and your sample balance.

How do you avoid leading screener questions?

Do not ask yes/no questions about the qualifying behaviour. Use a list of several categories or brands with a "None of these" option, randomise the order, and describe the survey topic in general terms so respondents cannot guess what gets them in.

What is a red herring question in a survey?

A red herring is a fake answer option, usually a brand or product that does not exist, placed among real ones. Anyone who claims to use it is likely answering carelessly or dishonestly to qualify, so they are flagged or removed.

What is a good incidence rate for a survey?

There is no universal good value; it depends on how niche the target is. General consumer targets often sit above 50%, while specific B2B roles or owners of one product can be in single digits. What matters is estimating it before launch, since required starts equal completes divided by incidence.

Should I screen my own customers?

Yes, but lightly. You already know they are customers, so focus on the criteria that matter for the decision, such as recent purchase or role. Tag them by source so you can compare them with a recruited audience in the same study.

Want a screener drafted from your objective, with quotas enforced live? SmartInterview is free for your first 50 respondents. See plans on the pricing page.

Table des matières

Articles associés

Inscrivez-vous gratuitement

Inscrivez-vous gratuitement

Inscrivez-vous gratuitement