Candidate Experience Net Promoter Score (cNPS): How to Measure, Benchmark and Improve It

TL;DR

• What it is: candidate experience net promoter score (cNPS) asks candidates how likely they are, from 0 to 10, to recommend applying to your organisation. cNPS = % promoters (9–10) minus % detractors (0–6), on a scale from −100 to +100.
• What “good” looks like: there is no independent benchmark. Published vendor figures range from −4 (North America, CandE 2025) to +17 (Starred, 2025 data) to +37 (Ashby customers, 2022–23), mostly because each dataset surveys a different mix of hired and rejected candidates. Compare yourself with yourself, by outcome and stage.
• Where the score falls: candidates rejected early score lowest in every dataset – as low as −23 for candidates rejected after an assessment. Hired candidates score around +80. One blended number hides both.
• How to run it: one 0–10 question plus one “why”, sent within days of the decision to rejected, withdrawn and hired candidates. Set a minimum number of responses before you report any segment.
• How to raise it: fix the early stages first. Decide faster, give rejected candidates feedback, and make the screening step feel fair and easy to complete. Then close the loop.

Most talent acquisition teams now treat candidate experience net promoter score as a key metric. Far fewer can say what their number actually means.

Two problems are common. Some programmes survey only candidates who reached interview, or only hires, so the score looks healthier than the hiring journey most applicants actually go through. Others blend every candidate into one number that nobody can act on. On top of that, most of the benchmark figures quoted online come from vendors, are several years old, or can’t be traced to a source at all.

This guide covers:

• the cNPS formula, with a worked example
• 2026 benchmarks, with their sources and biases shown
• how to run a cNPS survey that candidates actually answer
• five traps that make cNPS misleading
• a stage-by-stage plan to improve the score

A disclosure: Sapia.ai makes an AI interview tool used at the screening stage, which is where the benchmarks show cNPS is lowest. We’ve assessed 10M+ candidates (vendor-reported), so we’ve seen a lot of first-stage candidate feedback. We’ve tried to hold this guide to the standard we’d want applied to us.

What is candidate experience net promoter score (cNPS)?

Candidate experience net promoter score (cNPS) measures how likely candidates are to recommend applying to your organisation, on a 0–10 scale. It is the percentage of promoters (9–10) minus the percentage of detractors (0–6).

You’ll also see it called candidate net promoter score, candidate NPS or simply cNPS. Whatever the name, it’s a quick read on how candidates perceive your hiring process – and a proxy for what they’ll tell other people about it. Unhappy candidates spread negative word of mouth to friends, other job seekers and future candidates, and a strong employer brand is built as much by the people you turn down as by the people you hire.

The method comes from customer experience research. Fred Reichheld of Bain & Company introduced the Net Promoter Score in Harvard Business Review in 2003. cNPS asks the same “would you recommend us?” question of candidates instead of customers.

Wording varies by organisation and tool. Three real examples:

• “How likely are you to recommend a friend or colleague to apply for a job at {company_name}?” (Teamtailor)
• “How likely is it that you would recommend interviewing at [company] to a friend or colleague?” (Guide)
• “Based on your experience this far, how likely is it that you would recommend [company] to a friend or colleague?” (Jobylon)

“Recommend applying”, “recommend interviewing” and “recommend the company” measure slightly different things. Pick one wording and keep it fixed – tracking candidate NPS over time only works if the question doesn’t change.

cNPS vs eNPS vs candidate satisfaction

Candidate NPS measures intent to recommend; CSAT surveys measure satisfaction with the experience. They use different questions and are usually collected at different moments, so a satisfaction score out of 10 can’t be converted into a cNPS. Both are useful ways to measure candidate experience – just don’t report one as the other.

How to calculate candidate experience NPS (with a worked example)

The formula:

cNPS = % promoters minus % detractors

Promoters score 9–10 and detractors score 0–6. Passives (7–8) count towards the total number of responses but not towards either percentage. The result runs from −100 (everyone is a detractor) to +100 (everyone is a promoter). Report it as a whole number, not a percentage.

Here’s a quarter’s worth of responses for a fictional employer, split by outcome:

Across all 450 responses, 35.6% are promoters and 34.2% are detractors, giving a blended cNPS of +1.

That +1 looks neutral. It hides a hired-candidate score of +77 and an early-rejection score of −23. The early-rejection group is also the largest segment, with more than half the responses and 71% of all detractors. Always calculate and report cNPS by outcome and stage.

Watch your sample sizes. NPS swings a lot when only a few people answer. MeasuringU calculates that you need around 412 responses to show, with 95% confidence and 80% power, that a score beats a benchmark by 10 points (Lewis & Sauro, 2021). Recruiter-level or role-level cNPS built on a few dozen responses is mostly noise. Set a minimum number of responses before you report a segment (30–50 is a sensible floor), and show the number of responses next to every score.

What is a good candidate NPS score? 2026 benchmarks, and why they disagree

The honest answer first: there is no independent cNPS benchmark, and no industry standards for what counts as good. Every published set of candidate NPS benchmarks comes from a vendor’s own customers, or from a research programme a vendor owns. Here are the most credible ones, with their sources:

These figures are 40 points apart, and three things explain most of the gap:

1. Outcome mix. CandE deliberately weights not-hired candidates at 80% of its score. Most of Starred’s invitations go to rejected candidates. A dataset built mainly from candidates who reached interview will produce a much higher cNPS score than either.
2. Who answers. In Ashby’s candidate NPS data, response rates climb from 16% among candidates who didn’t get past the early stages to 43% among hires. Happier candidates are over-represented unless you account for it.
3. Region, sector and year. CandE reports North America separately, and Starred shows Tech & SaaS (+24) above its overall average.

So what is a good candidate NPS score? Benchmark against yourself, by segment, over time. If you need an external reference point, compare like with like (same outcome, same stage) and name the source and year. As a rough guide from the data above, hired candidates typically score between +50 and +85, and candidates rejected early usually score below zero. If your rejected candidates score above zero, that’s a strong candidate NPS by any of the published benchmarks.

You’ll see other numbers quoted elsewhere, such as “+20 is good” or “50+ is excellent”. The first has no published source. The second is Bain’s general guidance for a customer NPS score, not a candidate benchmark.

How to run a candidate experience NPS survey

The cNPS question often sits inside wider candidate experience surveys. Whether it stands alone or not, the same rules apply.

Who to survey

Survey every outcome at every stage of the recruitment process: candidates rejected at application, screen, assessment and interview, those who withdrew, and those you hired. Map the key stages of your process and tag every response with the furthest one the candidate reached, so you can report by stage later. The most common mistake when surveying candidates is to ask only interviewees or hires, which inflates the score and misses most of your candidates.

What to ask

Keep the candidate experience NPS survey to two to four questions:

1. The 0–10 candidate NPS question, with the wording kept fixed.
2. An open question: “What’s the main reason for your score?” This usually gives the most valuable insights, because it tells you why.
3. Optionally, one or two driver statements, such as “I was kept informed about my status” or “The process felt fair”.
4. Optionally, consent to be contacted about their answer.

Pull role, stage, location and recruiter from your ATS rather than asking candidates. Shorter surveys get slightly more candidates to answer. Ashby found a 20.1% response rate for surveys of up to 10 questions, against 18.2% for 11 or more (vendor data, 2022–23). For a wider question bank by stage, see our candidate experience survey sample questions.

When to send it

Send it soon after the decision. In Ashby’s data, 94% of responses arrived within nine days. Greenhouse lets teams send its survey between one day and four weeks after a hire or rejection, and some teams wait a few days so a recruiter can speak to the candidate first. If you survey at multiple stages, keep each survey short and limit how often anyone hears from you: Greenhouse won’t re-survey a candidate within six months, and Qualtrics recommends contacting candidates at most every 90 days at the apply and screen stages.

Which channel to use

Design for mobile, and use SMS as well as email for frontline and high-volume roles, where many job applicants apply from a phone.

Anonymity vs closing the loop

This is the real trade-off. In Ashby’s data, identified surveys got a 20.6% response rate, against 15.3% when anonymity was enforced. Anonymity can encourage more candid, accurate feedback, though. Guide, for example, holds responses back until the hiring decision is made. The catch is that you can’t call an anonymous detractor back.

A practical middle ground is to make the survey anonymous by default, add an opt-in “you can contact me about my answer” box, and close the loop at theme level for everyone else.

Privacy

This isn’t legal advice, so confirm your approach with counsel. Identified survey responses are personal data under GDPR and UK GDPR. The usual lawful basis is legitimate interests, and the survey should be mentioned in your candidate privacy notice. Many “anonymous” surveys are really pseudonymous, because the tool still holds the link to the candidate. Free-text comments and small segments can also identify people. Keep job alerts and other promotional content out of the survey, so it stays a piece of research rather than direct marketing under UK PECR and CAN-SPAM.

Choosing a tool

Start with your ATS’s built-in survey. The question that decides whether it’s enough is whether it can reach candidates rejected at application and screen, not just those who interviewed. Some built-in candidate surveys only fire after interview or after the final decision. If yours can’t reach early-stage candidates, you’ll need a dedicated candidate-feedback tool alongside it. We compare the main options in our guide to candidate experience software.

Reading your cNPS honestly – five traps

Five common traps make cNPS data look better or worse than it really is.

1. The blended score. One number hides the gap between hired candidates and early rejections, as the worked example shows.
2. Survivorship. Surveying only candidates who completed a stage, or only hires, inflates the score. That includes ratings collected at the end of an interview, because only the candidates who finished it can answer.
3. Skewed response rates. Hired candidates answer at nearly three times the rate of candidates who didn’t get past the early stages (Ashby: 43% against 16%). Report response rates beside your scores.
4. Small samples. Don’t rank recruiters or hiring managers on 20 responses. See the sample-size guidance above.
5. Targets tied to pay. If recruiters are paid on cNPS, expect selective sending and coached answers.

It’s also worth knowing that NPS has critics. An academic study (Keiningham et al., Journal of Marketing, 2007) couldn’t replicate the claim that NPS predicts business growth better than other measures. Treat cNPS as a useful way to track candidate sentiment, not as proof of better recruitment outcomes such as quality of hire.

How to improve candidate experience NPS, stage by stage

The benchmark data points one way: cNPS is won or lost early. Start with the lowest-scoring, highest-volume segment.

1. Candidates rejected at application, screen or assessment

The evidence. Starred’s 2024 data puts these candidates between −12 and −23, and CandE’s “not selected” candidates in North America scored −28 in 2025. CandE also found that 56% of candidates got no feedback after being rejected at screening or interview. Across its research, feedback raised candidates’ willingness to refer others by 50–75% (vendor-run benchmark).

What to do:

• Decide fast, and tell everyone. No silent rejections. Automate outcome messages so nobody slips through.
• Give timely feedback, even briefly. A short, specific note about what the role needed does more for sentiment than a generic “we’ve gone with other candidates”.
• Keep the application process short. Don’t ask candidates to retype what’s already on their CV or LinkedIn profile.
• Make the screening step feel fair and easy to complete. Ask job-related questions, don’t add time pressure the job doesn’t need, make it work on a phone, and don’t require a camera.

A rejection handled quickly and respectfully can still be a positive candidate experience, and it leaves a positive impression with someone who may apply again – or may already be a customer.

This is the stage Sapia works on. Our chat interview is untimed, text-based and mobile, and it scores candidates’ answers rather than their CV. Every candidate who completes it receives personalised feedback. At Woolworths, 82.6% of candidates completed the chat interview and rated it 9.2/10 for satisfaction (Woolworths case study, customer-reported). That’s a satisfaction rating from candidates who finished the interview, not a cNPS. More on that below.

2. Candidates rejected after interview

The interview experience shapes how finalists feel about you. Consistent, job-related questions across the interview process help candidates feel it was fair, even when the answer is no. CandE reports an interview-fairness NPS of 63 in North America in 2025 (vendor-run). To improve it:

• use structured interviews with the same questions and rating scale for every candidate
• tell candidates the format, length and who they’ll meet in advance
• brief every interviewer before they go in

Our guide to structured vs unstructured interviews covers the trade-offs.

3. Time in process, for every outcome

Delay hurts the candidates you reject most. In Ashby’s data, rejected candidates’ cNPS fell from 39 to 14 once a hiring process ran past 60 days, while hired candidates’ fell from 89 to 71. Set a decision deadline for every stage, and see our guide to reducing time to hire.

4. Candidates who withdrew

Withdrawn candidates often score well (+40 in Starred’s data). Survey them to find out why you lost them, not to fix the score.

5. Offers and hires

Scores here are already high, so don’t over-invest. Watch instead for drops between the final interview, the job offer and the start date.

6. Close the loop

Bain calls this the “inner loop”: the person closest to the experience – usually the recruiter or hiring manager – follows up quickly with unhappy candidates. Call back the detractors who opted in to contact, and share positive feedback with the hiring team too, so they know what to keep doing. For anonymous feedback, publish “you said, we did” changes on your careers site or in candidate emails. Asking for feedback matters in its own right: CandE found candidates invited to give feedback after the interview stage were 106% more likely to refer others (vendor-run, 2025).

For broader tactics across the whole candidate journey, see our guides to candidate experience and the candidate experience playbook.

Where Sapia fits (and where it doesn’t)

What Sapia isn’t: a cNPS survey tool. Sapia doesn’t send surveys triggered by your ATS after every decision, and it doesn’t offer an external cNPS benchmark. For measuring candidate NPS, use your ATS’s survey or a dedicated candidate-feedback tool.

What Sapia is: a way to improve the stage that drags cNPS down, which is the first screening interview. Sapia runs a mobile-first, untimed, text-first structured chat interview at the point of application, with an async video interview available as a later stage. Four things set it apart:

1. CV-blind scoring. The score rests on the candidate’s answers, not their CV, school or employers. That gives you a clear, job-related reason for every decision, which candidates can understand even when the answer is no.
2. An independent, published bias audit with a named auditor. BLDS ran 72 tests on our US and Canada models in late 2022 and found no evidence of practically significant disparate impact (FAIR framework and audit).
3. An untimed, text-based, mobile format. There’s no camera or voice, candidates have time to think, and it works on any phone.
4. Built-in detection of AI-generated answers, with published detection and false-positive figures (vendor-reported).

On top of that, every candidate who completes the interview receives personalised feedback, which addresses CandE’s biggest gap. Discover Insights reports candidate sentiment and drop-off alongside hiring data.

In practice (customer-reported): Woolworths candidates rated the chat interview 9.2/10 for satisfaction, with 82.6% completion, across its volume hiring. Spark NZ recorded an 8.9/10 candidate satisfaction rating from 16,000 interviews.

What those numbers are not: cNPS. They’re satisfaction ratings out of 10, given by candidates who completed the interview, before any decision was made. Don’t compare them with a post-decision cNPS. To see whether a better screening stage moves your cNPS, measure the early-rejection segment before and after the change.

Where we don’t fit, and what to choose instead:

• Spoken communication is the job, or you want a voice interview: HireVue launched a voice AI Interviewer in June 2026, though no independent audit of the voice product has been published.
• Published validity evidence is the deciding factor: SHL publishes deeper validity documentation than we do.
• Your main problem is scheduling speed: look at conversational scheduling tools such as Paradox, owned by Workday since October 2025.
• You need an ATS: Sapia isn’t one. It sits on top of the ATS you already run, through integrations.

What you should still ask us: our audit is from late 2022 and covers US and Canada models only, so ask us, and every vendor, for newer, role-by-role results. AI-text detectors have been shown to wrongly flag non-native English writers (Liang et al., 2023), and a false flag is a poor candidate experience in itself. Ask us for our false-positive rate for non-native writers.

If early rejections are pulling your cNPS down, a structured, CV-blind, untimed chat interview is one practical way to fix that stage at volume. See how Sapia’s chat interview works, or book a demo.

How to set up and improve your cNPS – a 5-step process

1. Define cNPS before you buy anything. Fix the question wording, the outcomes and stages you’ll report, and the minimum number of responses per segment.
2. Choose your survey tool by coverage first. If your ATS’s survey reaches candidates rejected at application and screen, start there. If not, add a tool that does.
3. Take a baseline for one quarter. Report cNPS scores by outcome and stage, with the response rate beside each one. Ignore segments below your minimum. Use the baseline to set realistic improvement goals for each segment – for example, moving early rejections from −20 to −10 – rather than one headline target.
4. Fix the lowest-scoring, highest-volume segment first. That’s usually early rejections. Pilot one change on one role family, such as faster decisions, feedback for rejected candidates or a lower-effort screening step. Then compare that segment’s cNPS and completion rate with your baseline. That’s how cNPS turns into data-driven improvements rather than a number on a dashboard.
5. Close the loop and report cNPS alongside how the recruiting process runs. Call back detractors who opted in, publish the themes, and report cNPS next to time-to-decision and the share of rejected candidates who received feedback.

Conclusion

Candidate experience net promoter score is simple to calculate and easy to misread. Track cNPS by outcome and stage, show response rates and sample sizes, and treat vendor benchmarks as a rough guide rather than a target.

Then spend your effort where the score is lowest, which in almost every dataset means candidates rejected early. Measure the score with a survey tool, and improve it with faster decisions, feedback for everyone and a screening step candidates find fair and easy to complete. If that screening step is your bottleneck at volume, see how Sapia’s chat interview works.

Net Promoter®, NPS® and the NPS-related emoticons are registered trademarks of Bain & Company, Inc., NICE Systems, Inc., and Fred Reichheld. Net Promoter Score℠ is a service mark of Bain & Company, Inc., NICE Systems, Inc., and Fred Reichheld.

FAQs

About Author

Barb Hyman

CEO & Founder