Candidate experience survey: questions and response coverage

Start with the question your team can act on

A candidate experience survey asks people about the recruiting process they went through. Use it to investigate communication, preparation or a difficult application step. Choose a process decision that someone on the team owns before collecting another set of opinions.

Choose one completed stage and one decision before you write the questions. For a screening call, you might want to know whether the candidate understood what would happen next. That gives the recruiting owner something specific to inspect: the closing message and the next update, rather than a broad instruction to improve the employer brand.

Download the editable questionnaire or the complete survey kit (ZIP). The kit contains a blank batch CSV, six worked cases, expected results, a local Python checker, a field dictionary and an action log. You can use the files with your own approved survey service and spreadsheet, without a Metix account. Nothing in the kit sends an invitation or collects a response.

This survey records the candidate's view of the process. An interview scorecard records the employer's assessment of the candidate. Keep the two jobs separate, including who can see the answers and how they are used.

Candidate experience survey questions to adapt

The starter questionnaire has common items and optional questions for an application, screen, interview or withdrawal. Pick only what belongs to the stage you are reviewing. Someone who never met an interviewer cannot tell you how prepared that interviewer was.

For Q1, the labels run from 1, Not at all clear, through Slightly clear, Moderately clear and Mostly clear, to 5, Completely clear. Offer Not applicable and Prefer not to answer as well. The worked example calls 4 or 5 positive and excludes the two non-substantive options from the item denominator. That is an editorial definition for this example, not a satisfaction standard. Record the threshold before looking at results.

Keep a version of the wording, answer options and invitation rule. A new wording or threshold needs a new version. Two charts labeled "clarity" can measure different things if one asks about the role and the other about the next interview.

SurveyMonkey's candidate experience guide offers question groups and a design process. A long menu is useful for choosing relevant items; it does not mean every candidate should receive every question.

Decide who receives the survey and when

Define the eligible group, the stage that triggers an invitation, the delay after that stage and the time allowed to respond. Record any exclusions and the reason for them. If you invite a sample, describe how you choose it. Sending only to people you hired gives you their experience, while leaving the experience of rejected or withdrawn candidates unobserved.

Ashby's survey design article discusses stage, timing and different survey configurations. Its observed response rates do not establish the best length, anonymity setting or send time for your team. This kit uses a seven-day follow-up period to demonstrate a consistent calculation; it does not recommend seven days for every hiring process.

Communicate the hiring result through your normal candidate process before a survey invitation could reveal it. Decide whether a person can receive surveys after multiple stages, and record that policy. A reminder belongs to the original invitation batch; counting it as a new invite inflates the denominator.

The introduction should explain the purpose, voluntary participation, who can read responses and how the feedback will be used. Let people skip questions. If you say feedback stays outside selection decisions, make sure the access and review process support that commitment. The downloadable introduction contains fields to replace, rather than a privacy promise you can paste without checking.

Check what anonymous means in your setup

A form without a name field can still collect identifying information. SurveyMonkey's official anonymity guidance explains that collector settings, tracking, integrations and the questions themselves affect what can identify a respondent. Check the actual collection route and describe it accurately.

Confidential access and anonymous collection are different promises. A small group or a recognizable comment can disclose an identity even without a direct identifier. An aggregate export does not remove that risk. Use the terms your actual setup supports, and check whether invitation tracking can be linked to answers.

Review the survey as a respondent before launch. Inspect the invitation, the form and an exported response. Restrict original comments to the people who need to handle them, then share themes or paraphrases with the wider team. An unusual event described in free text may identify its author even when you remove their name.

Show who answered before interpreting a percentage

In the first invented batch, twenty people receive a survey invitation after a screen. Eighteen have confirmed delivery. Six submit a response, five give a substantive Q1 answer and four choose Mostly clear or Completely clear.

The 80.0% describes five valid answers. It cannot describe the opinions of all twenty invitees. Fourteen people did not submit a response by the endpoint, and their views remain unknown. Delivery does not establish that someone read the invitation. A skipped item, Not applicable and Prefer not to answer also need their own handling instead of becoming a negative answer.

Define a submitted response in your collection system. Count each invitee once per batch, and keep the question denominator separate from the number of submitted surveys. Aggregate counts cannot reveal a duplicate person under two IDs; that reconciliation belongs in the underlying records.

Keep stage and outcome groups separate when you review coverage. The invented accepted-interview batch has eight responses from ten invitations; the declined-interview batch has two from fifteen. Those different response patterns make a single combined satisfaction percentage difficult to interpret. They do not establish why anyone accepted, declined or responded.

Run the local batch check

Each CSV row describes one question in an invitation batch sent on one calendar date under one stage, outcome definition and survey version. If you send on different days, make separate batches. Convert dates into one documented timezone first.

The example's first batch was sent October 1, 2026. Its seven-day window ends October 8, including that date. The November 1 report uses counts frozen through October 8. A more recent batch was sent October 28; its window ends November 4. Although six of its eight invitees have already responded positively, the checker withholds all percentages until that batch has its full follow-up period.

Download the local Python checker and the worked CSV into one folder, then run:

python3 candidate-experience-survey.py candidate-experience-example.csv --as-of 2026-11-01 > results.csv

Compare results.csv with the expected output. Start your own file from the blank CSV header and use the complete field dictionary. The script requires Python 3 and runs locally; the questionnaire and CSV files also work without running it.

For a closed batch, counts_through must equal its window endpoint. For an open batch it must equal the reporting cutoff. Refresh the counts when changing the cutoff; changing a date alone cannot supply missing observations. The script cannot verify whether your source export actually excluded late answers.

A blank delivered count leaves the delivered-response percentage unknown while the invitation rate can still be calculated. Blank answered and positive counts leave the item percentage unknown. When nobody answered, enter known zeros and leave the resulting percentage undefined. The checker returns a blank for a zero denominator instead of silently reporting 0%.

The script checks count order, duplicate batch/question pairs, conflicting metadata and date endpoints before returning a report. It does not connect to a survey platform, establish respondent uniqueness or authenticate an export. Prepare deduplicated, time-filtered counts in your own authorized collection system.

Give one supported change an owner

A comment about a missing update gives the recruiting owner a concrete check: read the message the candidate received and inspect when the next update went out. A small response count can still justify that investigation. It cannot tell you how common that problem is among everyone who took part in hiring.

The editable action log records stage, theme, the evidence, coverage limits, owner, decision, due date and verification. Its fictional example asks the recruiting owner to check whether the approved closing message names the next update. The next review should establish whether that message changed and whether the next comparable batch raises the same question.

Keep wording, audience, invitation rule and follow-up period comparable when reviewing later results. If several things changed, record them. A rising percentage alone does not show that a revised message caused the change. Avoid ranking recruiters from tiny groups with different roles, stages or outcomes.

Use recruitment funnel metrics to study progression between hiring stages. Survey response coverage asks a different question: whose reported experience you can actually see. Neither measure proves candidate quality or a fair outcome.

Use the right system for the next task

Keep survey collection and analysis in your approved survey service, ATS or other recruiting records. Use the kit for an independent calculation and a review conversation. It does not demonstrate a Metix survey feature, automatic export or integration.

If the feedback points to unclear role information, return to the hiring brief. If it points to an interviewer's preparation or how your team communicated a decision, the responsible people need to address that process.

Consider Metix when the issue is carrying an approved open role from sourcing through interview booking. Survey feedback about offer terms or interview decisions needs a different owner. In the current workflow, you define the brief, review matching evidence and approve outreach; Mira handles sourcing, interest screening and booking. Your team conducts interviews, issues offers and decides. Metix staff do not review self-serve shortlists. Staff-run Delivery is a separate service quoted on a demo.