What it is and what it's for#
Launching a survey is the easy part. What decides whether it is worth anything is what happens next: who can look at the answers, in how much detail, and down to what name.
Before looking at a single number there are two things to be clear about, because everything else depends on them.
The survey and the wave are not the same thing. The survey is the questionnaire: its questions and its configuration. The wave is each time you launch it. The same survey run in January and in June is two waves, and comparing those two snapshots is exactly what the distinction exists for. See Surveys, quizzes and climate.
A survey is either anonymous or it isn't, and that changes the screen. It isn't a filter you switch on when looking at the results: it is a property of the survey, decided when you create it. If it is anonymous, the results only exist in aggregate. If it isn't, on top of the aggregate you can see who answered what.
Anonymity is decided when you create the survey and cannot be changed once it has been launched. It is the only decision in the whole programme that admits no correction: if you promised anonymity and the survey was not anonymous, there is no way back.
How to set it up#
Results are not configured: they are calculated on their own from what people answer. What you do decide, and decide beforehand, is what you will be able to read.
- 1
When you create the survey, whether it is anonymous or named
This is the underlying decision. It determines which results screen you will have and what you can promise the people answering.
- 2
When you launch each wave, who it goes to
Locations, organisational units or tribes. The reach of the wave is what you will be able to slice afterwards.
- 3
When you look at the results, how you slice the data
The filters on the screen. Nothing changes here: you only choose which part you look at.
Scale questions are the only ones that produce an average comparable across waves. If you want to be able to say "we have gone from 3.4 to 3.8", that question has to be a scale question from the very first wave.
The filters#
Both results screens open with the same General filters block:
| Filter | What it does |
|---|---|
| Filter by Questionnaire (required) | Which survey you are looking at. Without this there is nothing to see. |
| Filter by Survey Wave | Which launch. With no wave selected, all of them are shown together. |
| Filter by Organisation | Narrows to particular organisational units. |
| Filter by Location | Narrows to particular sites. |
| Filter by Tribe | Narrows to particular groups. |
The screen starts empty: it is filled in with Refresh Data.
Combining Filter by Tribe with the organisation or location filters widens the population instead of narrowing it: anyone who meets one criterion or the other is included, not only those who meet both. If you want the exact cross of "this tribe at this site", don't get it from here.
How to read them#
What the respondent sees#
This is the part worth explaining properly, because people decide whether to answer based on what they believe happens to their answer.
- When there is a wave open for them, they see it with the points they get for completing it and the closing date.
- If the survey is anonymous, their answers cannot be attributed to them. It is recorded that they took part, because otherwise there would be no way to credit them the points.
- If the survey is not anonymous, their answers carry their name and their email address, and anyone with access to the results can read them one by one.
- Points are given for completing the survey, not for what is answered. No answer scores more than another.
"Anonymous" does not mean the same thing to you as it does to the person answering: people assume anonymous means that not even their participation is known. Say it plainly before the first wave — "we don't know what you answered, we do know that you answered, and that is only so we can give you the points" — and you save yourself the awkward conversation halfway through the campaign.
Anonymous results#
The Survey results block shows one card per question with the breakdown of answers. Scale and single-choice questions come out as a distribution; free-text ones, as a list of what people wrote, with no author.
There is no more depth than that, and it is deliberate: if the survey is anonymous, there is no per-person view to open.
In an anonymous survey, the free text is the only thing that explains the number. A 2.8 on "workload" doesn't tell you what to do; three comments mentioning the same shift do.
Named results#
When the survey is not anonymous, the same screen offers two modes:
The same as you would see in an anonymous survey: one card per question, with the average out of the top of the scale and the number of answers.
This is the view you always start with. It is for working out where to look.
You pick a person — the selector searches by first name, surname or email — and you see their answers, question by question.
It is for preparing a specific conversation, not for building a ranking.
The By employee view is the reason anonymity is not a decision to take lightly. If a named survey asks about climate, about the direct manager or about the intention to stay, that screen exists and someone is going to open it. Only ask by name what you would be willing to discuss with the person face to face.
The automatic report#
On anonymous results you can request an analysis generated automatically from the answers. It is not generated on its own: before doing so the screen warns you that the data is shared anonymously with an external provider, and you have to accept. The feature is marked beta and only appears if the project has it switched on.
The report summarises and organises what is already in the answers. It is useful for not having to read three hundred free-text answers by hand, not for replacing the reading of the ones that matter.
Taking it out#
Download exports the results to a spreadsheet. What comes out depends on the type of survey: in an anonymous one, the question and the answer; in a named one, also who each answer belongs to.
Frequently asked questions#
Can I compare two waves on the same screen?
There is no side-by-side comparison view. You look at one wave, note it down, look at the other. For long series, the practical thing is to export each wave and build the comparison outside.
If I don't select a wave, what am I looking at?
All the answers to that survey together, across every wave. It is an average of averages that is almost never what you want: if you have launched more than once, filter by wave.
Can I find out who hasn't answered?
In a named survey you can see who has answered, and work it out from there. In an anonymous one you can't, not even indirectly.
I changed a question between waves. Can it still be compared?
Not reliably. The series breaks: they are two different questions even if they look alike. If you need to ask something else, create a new survey and leave the original as it is.
Fewer answers appear than I expected.
Check the reach of the wave first: only the people it was sent to can answer. Then the dates: a wave only accepts answers while it is open.
Who can get into this screen?
Only accounts with admin permissions. A team manager does not see the survey results of their people. See The admin panel.
Limits and warnings#
- Anonymity cannot be changed once the survey has been launched. It is the decision with the most consequences in the whole module and it is taken before there is any data.
- An anonymous survey has no per-person view. It isn't hidden or restricted: it does not exist.
- The tribe filter and the organisation or location filters add up, they do not cross. Used at the same time they widen the population instead of narrowing it.
- The screen starts empty. If you see it blank, the most likely explanation is that the button that loads the data has not been pressed, not that there are no answers.
- With no wave selected, all of them are mixed together. Averages across several waves at once don't mean much.
- Some of the on-screen text is untranslated. It is cosmetic: the data is correct.
- A survey with few answers is not representative, and even less so if you slice it by site. With four answers in one shop, the percentage is anecdotal. See Analytics and reports.