Yes/No Survey Questions Are for Routing, Not Measuring
TL;DR
- A yes/no survey question, formally a dichotomous question, is a closed-ended question with exactly two answer options.
- It is the easiest question format to answer, which makes it the highest-response, lowest-information tool in survey design.
- Use it to screen, branch, confirm, and record facts. Do not use it to measure satisfaction, effort, or likelihood, where it flattens the differences you need.
- The practical rule: yes/no questions route respondents through a survey. Scales do the measuring once they arrive.
What is a yes/no survey question?
A yes/no question is a closed-ended question with two mutually exclusive answers. Survey methodology calls it a dichotomous question. "Did you contact support about this issue?" is one. "How satisfied were you with support?" is not, and should never be forced into one.
Because the answer requires no reflection and no typing, it carries the lowest effort of any question format. Pew Research Center's analysis of its American Trends Panel (92 open-ended questions across surveys run 2018 to 2021) found closed-ended questions are skipped by only 1 to 2 percent of respondents, while open-ended questions average around 18 percent nonresponse and can exceed 50 percent. Length still costs response, even at low effort: SurveyMonkey's analysis of 100,000 surveys (2009 to 2010 data, and the pattern is structural, not seasonal) found completion falls fastest across the first 15 questions. Every question added to a survey loses respondents, including the easy ones. A yes/no question therefore has to earn its slot by doing a job a scale cannot, not by being cheap.
Question formats sit in a small typology: open-ended questions collect free text, and closed-ended questions offer fixed options, which include multiple-choice, rating scales, and single-response formats. The yes/no question is the simplest single-response question there is: two options, one answer.
That is the trade. You get near-universal answers, and each answer carries one bit of information. The design question is never "are yes/no questions good", it is "which jobs deserve one bit".
A binary hides intensity. The customer who found you nearly impossible to deal with and the customer who found you slightly awkward both answer "no, it was not easy". One of them is about to leave, and your survey cannot see which.
When a yes/no question is the right tool
Four jobs, with examples you can lift.
Screening and qualifying. Establish who you are talking to before you measure anything.
- Do you currently hold a credit card with us?
- Have you visited a branch in the last 90 days?
- Are you the primary decision maker for this policy?
- Did you complete the application yourself?
Branching and routing. The answer decides which block the respondent sees next. This is the highest-value use, because it keeps every later question relevant.
- Did you contact support about this issue? (yes routes to the effort block, no skips it)
- Was your issue resolved on the first call?
- Did you use the mobile app for this transaction?
- Have you renewed this policy before?
Consent and follow-up permission.
- May we contact you about your feedback?
- May we share your comments with the branch team?
- Would you like a copy of your responses?
Factual states. Things that genuinely are binary.
- Do you currently use auto-pay?
- Is this your primary account?
- Did the card arrive within 7 working days?
- Have you used the rewards portal?
When it is the wrong tool
Anything you intend to measure has a distribution, and a binary destroys it. Three common misuses, with the fix:
Wrong as a binary | Use instead | Why |
|---|---|---|
Were you satisfied with the service? | A 5-point satisfaction scale | "Yes" merges the delighted with the barely retained. The difference between them is the finding. |
Was it easy to resolve your issue? | An effort scale scored with the net method, which Numr calls the Net Easy Score | A net score (percent easy minus percent difficult) exists to keep the extremes visible. A binary buries exactly the customers most likely to act. |
Would you recommend us? | The 0-10 NPS question | The promoter, passive, detractor split is the instrument. A yes/no cannot produce it. |
The pattern in all three: measurement questions need the ends of a scale to be visible, because the ends are where behaviour changes. Routing questions do not, because their only job is a fork.
Writing rules that keep the answers honest
One fact per question. "Was the branch staff friendly and fast?" is a double-barreled question. A respondent whose teller was friendly and slow has no truthful answer. Split it, or scale it.
Keep the wording neutral. "Did our award-winning support team resolve your issue?" leads the witness. "Was your issue resolved?" does not.
Only ask what memory can answer. "Did you receive our email in March?" fails not because it is binary but because nobody knows. Binary questions should be answerable instantly or they inherit the skip rates of hard questions.
Give an honest out where the world is not binary. If a meaningful share of respondents genuinely cannot say yes or no, add "not sure". If you find yourself needing "sometimes", the question wants a scale, not a third option.
Put gating questions early. Screening and branching questions belong at the front, where they cost seconds and save every respondent the questions that do not apply to them.
The routing pattern in practice
The most useful yes/no question in a CX programme is the one that triggers work. "Was your issue resolved?" answered "no" should not become a data point. It should open an action: an owner, a follow-up, a closure. That is the difference between a survey that reports and a closed-loop programme that recovers customers.
Short, low-effort formats are also what make high response rates possible in the first place. In one Numr insurance engagement, moving from sampled phone surveys reaching under 1 percent of renewing customers to short digital surveys at the point of interaction, a single scored question with targeted follow-ups, took the programme to more than 1,200 responses a day. Colt cut its enterprise survey to three elements for the same reason: in enterprise accounts, every additional question is one more reason to close the tab.
Amitayu Basu, CEO and Co-founder, Numr"The yes/no question is the cheapest question you can ask, so spend it where it works: deciding what happens next. If the answer 'no' does not trigger anything in your organisation, you did not need the question."
Frequently asked questions
Is a yes/no question the same as a dichotomous question?
Yes. Dichotomous is the formal term for any two-option question. Yes/no is the most common form; true/false and agree/disagree pairs are the same structure.
Do yes/no questions increase response rates?
They have the lowest skip rates of any format, 1 to 2 percent on closed-ended questions in Pew's panel data versus around 18 percent for open-ended questions. A survey built on low-effort questions completes more often, but the gain comes from overall length and effort, not from converting measurement questions into binaries.
How many yes/no questions should a survey have?
As many as have routing or screening jobs to do, and no more. They cost little individually, but they still count toward total length, and completion falls fastest over the first 15 questions.
Can the NPS question be asked as yes/no?
No. The Net Promoter Score requires the 0 to 10 scale, because the score is the percentage of promoters (9 to 10) minus the percentage of detractors (0 to 6). A yes/no follow-up after the NPS question, such as "May we contact you about your experience?", is a good use of the format.
Should I add a "not sure" option?
Only when a meaningful share of respondents genuinely cannot answer. For factual questions about their own behaviour, a forced binary is usually honest. For anything resembling a judgment, a scale serves you better than a third option.
Net Promoter Score and NPS are registered trademarks of Bain & Company, Fred Reichheld, and Satmetrix.