Employee survey: the questions that give answers

Fewer questions give better answers

It is tempting to ask about everything. But every question costs time, and fatigue gives worse answers towards the end. More importantly: every question quietly promises that someone will act on the answer. Only ask what you are willing to follow up.

A main survey of around 30 statements takes about four minutes to answer. Short pulses between main surveys should take under a minute.

The topics to include

Section 4-3 of the Working Environment Act and chapter 1A of the regulation on the performance of work point to the factors that matter most for the psychosocial working environment. A survey used for mapping should cover them:

Eleven example questions

The questions work best as statements people take a position on, on a five-point scale: strongly disagree, somewhat disagree, neither, somewhat agree, strongly agree. Here is one statement per topic, as they appear in Orgpuls (translated):

1. I can speak up about unacceptable conditions without fearing the consequences
2. My workload is manageable over time
3. I rarely get tasks without knowing what should be deprioritised
4. I can be in contact with colleagues during the working day
5. I am rarely left alone in emotionally demanding situations
6. I get support from my manager when the job gets demanding
7. I have a say in how I carry out my work
8. I am treated with respect whoever I am
9. I know what is expected of me in my job
10. My colleagues step in when I need help
11. The work I do matters

Questions outside the index

Some questions do not fit an overall index, but are worth asking:

• Recommendation: «How likely is it that you would recommend us as a workplace to someone you know?» Reported as a figure of its own.
• Offensive behaviour: «Have you experienced or witnessed offensive behaviour, harassment or bullying at work in the last twelve months?» Should only be shown as a count for the whole organisation, never per group.

An open comment field often gives the most important findings. But free text can reveal who wrote it, even when no name is stored. Treat comments with the same care as figures for small groups.

Anonymity: why a limit of five answers

People answer honestly when they know the answer cannot be traced back. That takes more than not storing the name: a department of three people is recognisable regardless. So no group should be shown until a minimum has answered. In Orgpuls the limit is five, and it cannot be lowered. Read more in Anonymous employee survey: what does anonymous mean?

What to do with the answers

The result is not the answer – it is the start of a conversation. Share it with employees within two weeks. Choose the two or three findings that matter most and turn them into measures with an owner and a deadline. Do not take on nine things at once.

How to read the result

In Orgpuls the answers are converted to an index from 0 to 100: strongly disagree becomes 0, strongly agree becomes 100, and each factor gets the average of its three statements. A factor at 65 or above counts as low risk, 50–64 as medium and below 50 as high.

The level is still less interesting than the movement. A factor at 61 says little on its own. A fall from 64 to 61 since last year, while the rest stand still, is a signal. Look at the factors before the total – the total can stand still while two factors move in opposite directions.

Why five answer options

A five-point scale with a midpoint lets people say «neither» when that is what they mean. Without a midpoint they are forced to take a side, and that adds noise. More steps rarely give better data – the difference between 6 and 7 on a ten-point scale means different things to different people.

Then measure again on the factors where measures are running. You see whether anything has moved, and employees see that their answers led somewhere. That is what keeps the response rate up next time.