Getting the best from research: Surveys

I’ve blogged recently about focus groups and interviews. The other most used research method of mine is surveys. And they’ve been keeping me very busy over the last few weeks. The design, the build (with no little help from Survey Monkey) and, most of all, the reporting and analysis.

The great advantage of a survey is to quickly get the opinions of as many people as possible. Unless you’re planning a schedule of pop-up surveys, this is your one chance for the maximum information. But get tempted to ask too much and your recipients switch off. Unless they’re 100% convinced that they’ll see action off the back of your survey – and who can guarantee that? – too long or too complex a survey will see them answering without due consideration, not expanding on answers, or simply abandoning.

So, what’s to do? Well the best place to start is at the end.

If you know what you really, really want to get out of the survey, then you can focus on that, and start to be brutal about what you cut. Yes, you can get still get plenty of other information out too. And some clever question and page design can improve the flow and how long the survey feels. 

And that focus will help in the analysis. Even with just a handful of questions with a few options each, there’s already a lot (tell me if the maths is too complex…) of ways you could cut that data. The other temptation is to start asking “How did female middle managers who replied on a Wednesday compare?” It’s unlikely that that’s where the big story is. If there is another big story, and your analyst is worth their salt, they’ll find it.

A couple more points:

• Free-response questions. Some of the big employee surveys include a couple, and then provide no analysis on them. Crackers. That’s where the context is. Throw in the questions, be careful about whether you require people to answer them, then get them analysed.
• Following up: Of course, you can’t interrogate people’s answer in a survey. So get their permission to follow up, to perhaps come to focus group, be interviewed or have a handful of more precise questions put to them.

Then you’ll get genuinely valuable insight, and not just a benchmark. Of which more soon…