Short answer
An employee survey response rate describes participation in a defined campaign. For a known eligible invitation list, divide completed responses by eligible people invited and multiply by 100. State the completion rule, keep partial responses visible, and check participation across relevant workforce groups. The percentage alone cannot establish representativeness, honesty, or the reason somebody did not answer.
Fix the denominator before comparing results
Start with a roster reconciled to the campaign's opening date. Record how people on leave, new starters, leavers, duplicate invitations, and undeliverable invitations are handled. A bounced email may reveal an access failure; it is not automatically a reason to remove an eligible employee from the denominator.
Use a small campaign record:
| Field | Decision to document |
|---|---|
| Eligible population | Who the campaign is intended to include |
| Invited population | Eligible people actually invited, with exclusions explained |
| Complete response | The rule chosen before collection |
| Partial response | Started or submitted some answers without meeting that rule |
| Question denominator | People providing a usable answer to the specific question |
| Field period | Opening, closing, and reminder dates |
| Delivery problems | Unreachable addresses or inaccessible channels requiring follow-up |
For a simple known-population campaign, report completed responses / eligible people invited × 100. Do not present this practical internal metric as a substitute for a formal survey standard when eligibility or sampling is more complex. AAPOR's Standard Definitions explains why case dispositions and a consistent calculation matter. It also makes clear that response rate alone does not establish the presence or size of nonresponse error.
A fictional example: the average hides the access question
Suppose a fictional employer invites 200 eligible employees. It receives 120 complete responses, 20 partial responses, and no response from 60 people. The complete-response rate is 60%. The partial responses remain a separate category; adding them without explanation would change what the reported percentage means.
| Population | Eligible invited | Complete | Complete-response rate |
|---|---|---|---|
| Office | 50 | 45 | 90% |
| Stores | 100 | 55 | 55% |
| Night shift | 50 | 20 | 40% |
| Total | 200 | 120 | 60% |
This does not prove that night-shift employees are less engaged or less trusting. It identifies a question to examine: did those employees receive an accessible invitation, have a suitable time to participate, and understand the purpose? Their answers may also differ from respondents on topics that cannot be observed from the roster.
If only 100 of the 120 complete respondents answer the workload question, use 100 as that question's response denominator. Report missing answers separately from a neutral response or a decision to skip.
Diagnose before sending another reminder
Choose a change from the observed friction, then test whether it helps.
| Observation | What to check | A bounded next step |
|---|---|---|
| Invitations do not reach a worker group | Channel access and roster quality | Test an approved accessible channel with that group |
| People start but stop at one question | Wording, technical behaviour, sensitivity | Pretest the question and repair a confirmed obstacle |
| One language has more incomplete responses | Translation and meaning in context | Review wording with people from that population |
| Employees cannot explain how answers will be used | Invitation and reporting explanation | Clarify the decision, audience, and confidentiality limits |
| The same topic returns after several campaigns | Ownership and prior follow-through | Publish the decision or next review before asking again |
AAPOR's survey best practices recommends pretesting with people similar to the intended respondents, piloting the procedures, keeping burden manageable, and distinguishing missing answers from answer options. Apply those principles to the actual devices, languages, shifts, and topics in your campaign.
Do not ask managers to identify or pressure nonrespondents from an anonymous reporting view. Use the participation and access information the process legitimately provides. A higher number is not a useful outcome if obtaining it changes the promise made to employees.
Use benchmarks as context, not an acceptance test
Before comparing your rate with a vendor report or another employer, ask whether the figures use the same population, invitation method, completion rule, question set, period, and survey type. A pulse on one operational question and an annual questionnaire are different collection tasks.
Set your next-campaign target around a diagnosed issue: make an invitation reachable, reduce unexplained drop-off, or include a previously missing worker group. Keep the measurement definition stable. If you change it, show the old and new definitions rather than presenting an artificial improvement.
Keep survey completion separate from review follow-through
A completed questionnaire, a performance conversation that took place and an agreed follow-up are different events. Do not use one percentage for all three. For a review process, record the people whose review was due, the meetings held under that rule, the written records completed and the actions agreed, each with its own definition and period. A submitted form does not prove that a useful conversation happened; an undocumented meeting is not automatically proof that no conversation occurred.
If the immediate problem is organising and recording those meetings, use the performance-conversation preparation and record pack. Keep the participation calculation tied to the event it actually counts, then examine the quality and follow-through separately.
Make the report useful to the decision owner
A concise readout includes:
- eligible, invited, complete, partial, and question-level counts;
- relevant participation differences, with small-group protections;
- known access or collection problems;
- what respondents described and what remains unknown;
- the decision this evidence can reasonably support;
- an owner and review date for the next step.
For open comments, use the qualitative engagement analysis guide. For the collection process across channels, use the employee voice analytics guide.
Explore a conversation when you need more context
If the survey identifies a work question that needs examples rather than another rating, explore Lontra's employee conversations and manager briefs. Start with a focused question, then examine whether the responses help prepare a useful human discussion.
Track participation and missing groups in that format too. Choosing a conversation does not remove selection effects or establish that it will achieve a higher response rate. Its value should be judged by the context obtained and the decision it helps the team examine.