Candidate Experience Survey Template
A 12-question candidate experience survey you can answer right now. Send it once, after someone has left your hiring process; which of three exits they took decides the block on the end. Read the candidate experience survey questions below, copy them, or open the template in the editor.
- 12questions
- About 3 minto complete
- One send2 to 4 days after the outcome
- 3 versionsnot selected, withdrew, hired
Which of three exits a candidate took decides what you send them
Not a date on a calendar. The moment that matters is the one where the person stopped being a candidate, and how that happened changes who answers and what their answer is worth.
Send it once, after they have left the process. Not after each stage, and never while a decision is open. Starred, the hiring-feedback vendor with the largest openly published dataset on this subject, advises one survey per candidate a few days after the process ends: someone asked while still in the running gives biased answers, because people assume warm feedback helps their chances.[4] A stage-by-stage design collects its best scores from the people with the most to lose.
Almost everyone you write to was turned down. In 2025, 89.7% of candidate survey invitations on that platform went to rejected candidates, most of them at the application stage.[1] The long-running industry benchmark programme targets over 80 percent non-hired candidates too.[8] The question is not whether to survey rejected people, but what happens to your number when they are most of it.
Two hand-offs first. The same process rated from the inside is the hiring manager recruitment survey. If the person has already started, their first days belong to the new hire orientation survey.
| How it ended | Send it | Add this block | What comes back |
|---|---|---|---|
| You did not select them | 2 to 4 days after they are told | Was the reason clear, would they apply again | Almost all your sends, about a fifth answer[2] |
| They withdrew | 2 to 4 days after they tell you | What made them stop taking part | Smallest group, most even answers[2] |
| You hired them | After they accept, before day one | Does the job so far match the advert | Half answer, and hold criticism back[2] |
One questionnaire, three reporting lines. About 21 percent of rejected candidates answer, 22 percent of those who withdrew and 50 percent of the people you hired.[2] Groups that different cannot share a number, though they can share the core twelve; only the last two or three items change. If one link is all you can manage, question 1 records the exit in the candidate's own words, so a single send still splits cleanly.
Why this candidate experience survey has to work for someone who only sent a form
Most of the people you are writing to never spoke to anyone at your company. A question about interviewers is dead weight for them, and dead weight is what makes a survey feel like a formality.
What is deliberately missing. No name, no email, no reference number, no demographic band. The people you most need to hear from have just been turned down by you, and an identity field is where that conversation ends.
All 12 are already built, so you can open them in the questionnaire maker and change any wording first.
The candidate experience survey questions, with a block for each exit
Every question, its scale, and what a poor answer to it looks like. The last four blocks are the endings: three by exit, one for a test or a task.
Where it ended, and how far it got
Two taps that decide how every other answer is read. Ask them even when your records already say so.
- How did your application with us end?The split you report by. Skip it and your number moves with your hire rate.
- What was the last stage you reached?A rejection after one form is not the same event as one after four interviews.
What you promised, and what you asked of them
All three work for someone who only filled in a form, which is most of your list.
- The advert gave me an accurate picture of the job.A low score reappears later as a new hire saying the job was not what they read.
- It was clear what the hiring process would involve and how long it would take.Disagreement means nobody described the process. The cheapest thing here to fix.
- The time the application asked of me was reasonable for this job.A long form for a screening round is the version people resent.
How they were treated while you decided
Whether anyone spoke to them, how, and whether anything asked was none of your business.
- I was kept informed while my application was being considered.The item people who walked away score lowest. Silence is what they leave over.
- Everyone I dealt with treated me with respect.A low score is about one named person, and the open answer usually names them.
- Nothing I was asked felt intrusive or unrelated to the job.The fairness check that never asks about a protected characteristic.
The verdict, as a number and in their words
One judgement, one checklist you can count, one open answer, then the closing rating.
- Looking back, this was a fair way to judge someone for this job.Read by exit: a rejection drags this one down on its own.
- Did any of these put you off along the way?Read as counts. The top item is your first fix, whatever the scores say.
- What is the one thing you would change about how we ran this?Asking for one change returns a sentence. Asking for comments returns a blank.
- How likely are you to recommend applying here to a friend or colleague?Percent at 9 or 10 minus percent at 0 to 6, worked out once per exit.
Add this block: you did not select them
Paste these on the end of the copy for candidates you turned down. Nobody else sees them.
- The reason I was not selected was made clear to me.This group's biggest complaint, and it is about the message, not the decision.
- Would you apply for another job here?A no from someone you interviewed costs a future applicant and their referrals.
Add this block: they withdrew
The smallest group, the one almost nobody asks, and the one that answers most evenly.
- What made you stop taking part?Answers cluster on the two communication options, and both are yours to fix.
- Was there a point where we could have changed your mind?The nearest thing to a list of the offers you never got to make.
Add this block: you hired them
Send it before the first day and keep it to what the process promised.
- What I have been told about the job so far matches the advert I applied to.Disagreement before someone even starts is the earliest warning you get.
- Was there a point where you nearly dropped out?Names the leak that cost you the people who did drop out.
Swap in: you use a test or a task
Two items for the stage that produces the angriest answers of all.
- The test was a fair sample of the work this job involves.A low score plus a rejection at that stage is your worst combination.
- I knew before I started how long the test would take.An unbounded task is the one candidates describe as unpaid work.
Where the wording comes from. No published, licensed scale exists behind this subject, so every line is written for this page and free to edit or cut. One caution: keep the core twelve fixed once you have sent them, because a reworded question starts a new trend line. Everything true of any survey, from the link to the closing date, is in the survey research guide.
One number, three populations, and why you must never average them
The recommend question gives you percent scoring 9 or 10 minus percent scoring 0 to 6. Work it out three times, once per exit, and publish it three times.
A blended score is a rejection score wearing a disguise. Nine in ten invitations go to people you turned down, so a company-wide number is mostly their number. It also moves for the wrong reason: hire more people next quarter and the blend rises on its own, with nothing about the process having changed.
Reference points, and what they are. The table is that platform's 2024 averages by exit, not an industry benchmark and not a target.[2] Its 2025 figures move the rejected average a point lower, to -7.[1] Read the table for the distance between the groups, then compare yourself only with yourself.
| Who answered | Their 2024 average | What it tells you |
|---|---|---|
| You hired them | +82 | Gratitude and a new job are doing the work. |
| They withdrew | +40 | The most honest number, from the smallest sample. |
| You did not select them | -6 | What your blended score is actually made of.[2] |
| Turned down after an assessment | -23 | The worst exit: unpaid work, then a no. |
| Turned down after an interview | +1 | The most willing rejected group, and the least bitter. |
The pairing nobody prints. Candidates turned down after an interview answer most willingly and complain least. Those turned down after an assessment are the most negative of all.[2][3] Your angriest answers come from the stage where you asked for work and gave back a no.
Do not read it on day one. On that platform the running score starts high, swings negative, then settles about halfway between.[4] The strongest reactions arrive first, so the late answers are the representative ones. Time works on the judgement itself as well: pooling studies that measured the same applicants twice, the drop in how fairly they rated a process was smaller, sometimes absent, when more time had passed between the last assessment and the decision.[5]
Some of your numbers will be too small to publish. Hire ten people a year and the hired copy returns about five answers in twelve months. Under thirty answers in a group, report the tick-all counts and the open answers, and say the score is not ready.
What these answers cannot tell you
The group handing you most of the data has a reason to mark you down, and that is worth knowing before you read a single score.
One line about the law, in the United States. The ADA bars disability-related questions of an applicant before a conditional offer,[6] so this template carries no health, disability or adjustment item and no block adds one.
The incentive problem, admitted by the people publishing the data. A rejected candidate may mark you down for not advancing rather than for anything you did, and a hired one may hold criticism back. The most balanced answers come from people who withdrew, the group almost nobody surveys.[2] Rejected answers are still your only evidence about the stages people leave at.
What the hired answers are, and are not. New hires who call their candidate experience exceptional are twice as likely to say their responsibilities match what recruitment promised.[7] So the hired block is an early reading on the first year, not an audit; it asks about the advert, not the interviewers.
One named interviewer is a different conversation. If the same person keeps appearing in the open answers, the manager feedback survey asks their team rather than their candidates.
Two neighbours, so nothing is asked twice. A former manager rating the person you are about to hire is the skill reference check. Measuring the skill itself is the skills assessment survey.
Candidate experience survey FAQ
What is a candidate experience survey?
A short survey sent to someone who has been through your hiring process, asking what it was like from the outside. It goes out once they have left it, whether you turned them down, they withdrew or you hired them.
Should we survey candidates we did not hire?
Yes, and they are most of the point. They are the only people who can describe the stages candidates leave at, and they are almost your whole list. Report their answers apart from the ones you hired.
What if we only want feedback from people we interviewed?
That is usually what an interview experience survey means: the same instrument, sent to one slice of the funnel. Question 2 records the last stage reached, so you can filter to those people and still keep everyone who left earlier.
Should the survey be anonymous?
Ask for no name and no email, which is what this template does, then be accurate about the rest: whoever owns the survey can open every response as a row. Give it an owner outside the hiring panel, and name them in the invitation.
How long should a candidate experience survey be?
Twelve questions and about three minutes, plus two or three in the block on the end. Every question you add is asked of someone you just turned down.
Methods and sources
What this template is based on
Question wording is original; no published scale exists behind this subject. The send rule, the exit split and the scores by exit are Starred's own reports plus the industry benchmark programme;[1][2][3][4][8] how time changes an applicant's judgement of fairness is peer reviewed;[5] the applicant boundary is US federal guidance.[6]
Question licensing
Nothing licensed is reproduced. The 0 to 10 recommend wording is free to use; cNPS is descriptive. Net Promoter, NPS and Net Promoter Score are trademarks and service marks of Bain and Company, Inc., NICE Systems, Inc., and Fred Reichheld;[10] SuperSurvey holds no licence or affiliation.
Limits and disclosure
Every figure here is Starred's own platform data, not audited and not a probability sample. The two to four day window is our judgement, and a low response rate is not by itself proof of bias.[9] The ADA line is US law only; CandE findings behind a form are not cited.
SuperSurvey makes the editor this template opens in; everything here works without it.
References
- Starred. 2026 Hiring Benchmarks Report. Starred, 2026. starred.com
- Starred. 2024 Candidate Benchmark Report. Starred, 2025. starred.com (PDF)
- Starred. "Candidate NPS benchmarks: what are they?" Starred blog. starred.com
- Starred. 2023 Candidate Experience Benchmark Report. Starred, 2024. starred.com (PDF)
- Konradt, U., Oldeweme, M., Krys, S. and Otte, K.-P. "A meta-analysis of change in applicants' perceptions of fairness." International Journal of Selection and Assessment 28, no. 4 (2020): 365-382. d-nb.info (PDF)
- U.S. Equal Employment Opportunity Commission. Enforcement Guidance on Disability-Related Inquiries and Medical Examinations of Employees under the ADA. EEOC, 2000. eeoc.gov
- Kemp, A. "The Lasting Impact of Exceptional Candidate Experiences." Gallup, 16 October 2024. gallup.com
- Survale. The CandE Benchmark Research and Award Program. Survale. survale.com
- Groves, R.M. and Peytcheva, E. "The Impact of Nonresponse Rates on Nonresponse Bias: A Meta-Analysis." Public Opinion Quarterly 72, no. 2 (2008): 167-189. academic.oup.com
- Bain and Company. Trademarks and licensing. Net Promoter System. netpromotersystem.com
Send it to the next candidate who leaves your process
Twelve questions, about three minutes, no name asked. Open it in the editor, add the block for the exit they took, and send it once.
Use this template