The most damaging thing a company can do with an engagement survey is run it, share nothing, and change nothing. That teaches people not to answer the next one honestly.
- ›Designing a survey worth answering
- ›The anonymity that makes it honest
- ›What to do with results
- ›Cadence
Designing it
Short. Ten to fifteen questions for a full survey, three to five for a pulse. Long surveys produce declining quality toward the end and declining response rates over time.
Specific and actionable. "I have the tools I need to do my job well" produces a fixable finding. "I am satisfied with my employment" produces a number.
Consistent across runs. Changing questions each time makes trends impossible, and trends are the point.
One open text field. The structured questions tell you what; the open field tells you why, and it is usually the most valuable part.
Anonymity that is real
This determines whether answers are honest.
- ›Minimum group sizes enforced in reporting, typically five or more, so small teams cannot be identified.
- ›No demographic cross-tabs that would reconstruct an identity.
- ›Never attempt to identify a respondent. One incident, discovered, ends honest answering permanently.
- ›Say what is anonymous and what is not, plainly, before people answer.
Important: If a manager asks who wrote a negative comment, the answer is no, and the reason to hold that line is that the entire instrument depends on it.
What to do with results
The follow-through determines whether the next survey is worth running.
Cadence
Annual or semi-annual full survey, plus monthly or quarterly pulses on a few questions.
Weekly surveys produce fatigue and falling response rates, which then look like falling engagement. Resist the tooling that encourages it.
Reading the results
Trends over levels. A 6.2 that was 5.4 last time is a different situation than a 6.2 that was 7.1.
Variance by team. Company averages hide the teams where the experience is genuinely bad, and those teams are where the actionable finding is.
The open text. Read all of it. Themes in free text are usually more actionable than the scores.
FAQ
What response rate is good?
Above 70% is healthy for a full survey. Falling response rates are themselves a signal, usually that previous surveys changed nothing.
Should managers see their team's results?
Yes, with minimum group sizes enforced, and paired with support to act on them. Results delivered without help produce defensiveness.
What if results are bad?
Treat it as information rather than a messaging problem. Engagement scores respond to changes in the experience, not to communication about the experience.
Brainis Engagement OS includes pulse surveys, eNPS, and exit interviews with enforced minimum group sizes. See it in the help center.
Sharing insights on business operations, AI, and modern team management.
Run your company on Brainis
All 11 Operating Systems on every plan, from $29 a month. No per-seat pricing — you pay for AI capacity, not headcount.
See pricing