A reply establishes what that person wrote about the named question. It does not establish the view of staff who did not reply. Omitting a name does not make the file anonymous. Write a response rate only when invited and replied counts are valid whole numbers for the same roster and replies do not exceed invitations. Keep identifying free text out of the shared tally.

Use the engagement guide after the replies are tallied. Record journey distances on the travel survey. Keep a staff tally apart from a materiality assessment. Check people-team duties before asking extra personal questions.

What a reply can establish

The ICO introduction to anonymisation says information is anonymous only when a person is not identifiable, including when the data is combined with other information. That page is under review because of the Data (Use and Access) Act. Pseudonymised replies remain personal data.

This template does not establish anonymity. That is different from saying a staff survey can never be anonymous. identifiability_status is not_assessed, possibly_identifiable or personal_data. anonymous is not an allowed value. If identifying detail is already known, do not leave the status not_assessed.

Do not ask special-category questions such as health, political opinions or trade-union membership. Do not ask commuting distances. There is no engagement score. For commuting emissions, use the separate commuting survey, which records journey legs and has its own privacy notes.

Ask six questions and tally the replies

Download the questionnaire people read and the tally used for counts.

The questionnaire is plain text. Before you send it, fill the operator preparation note: purpose, that participation is voluntary, who can see the responses, and the retention arrangement you have chosen. Do not invent a legal basis, a universal retention period, or a promise of confidentiality.

The six questions are:

  1. Have you seen a named workplace sustainability priority? yes / no / not_sure
  2. Name one barrier in your own work, or write none_stated
  3. Name one action you could take in your role, or write none_stated
  4. Do you know who to ask? yes / no / not_sure
  5. Optional role group, only if a colleague could not identify the speaker; otherwise leave blank
  6. Anything else about workplace practice, or write none_stated

The tally CSV uses record_type survey or question. A survey row holds survey_id, invited_count, reply_count, response_rate, response_rate_unit, identifiability_status, suppression_note and operator_note. A question row holds survey_id, question_id, yes_count, no_count, not_sure_count, missing_count and free_text_status. Question rows do not repeat the invited or reply counts. Do not add those header totals together as extra replies.

invited_count and reply_count are non-negative whole counts for the same survey and roster. Count one response per invited person. Under that method, replies cannot exceed invited people. The rate stays blank if either count is missing, invitations are zero, a count is negative or fractional, replies exceed invitations, or the eligible-response basis is unknown. Write the rate to one decimal place with the unit percent. 8 of 12 is 66.7 percent.

Question counts must not exceed eligible replies. An unanswered question stays missing, not no. Keep identifiable free text out of the public tally.

Worked example: eight replies, no anonymity claim

The example is fictional. Northgate Services invited 12 people and received 8 replies in March 2026.

Record What was written Status
Survey header Invited 12, replied 8. Rate 66.7 percent. Operator note names purpose, voluntary participation, who can see the file, and retention until 12 March 2027. Rate written because both counts are valid. Not a workforce result.
Identifiability A private barrier note could identify a person. That wording is not in this tally. possibly_identifiable. Not left not_assessed.
Q1 yes 5, no 2, not_sure 1, missing 0. What those eight people wrote.
Q5 Role group left blank. missing 8. No role breakout in a twelve-person office.

If invited count is missing, zero, fractional or smaller than replies, leave response_rate blank. If Q1 yes, no and not_sure add to more than eight, the question counts are not usable.

Limits

Do not say the survey is anonymous, confidential-by-default, statistically representative, or an engagement score. The ICO page is guidance under review, not a completed legal opinion for this firm.

Your next action

Use the engagement guide after the replies are tallied.

Sources