- Keep your survey short: a long survey is less likely to be taken in full, or at all, and my guess is that the longer a survey is, the more skewed the resultant data will be, because only those with really strong opinions (positive or negative) will slog on through it all. The longer your survey is, the less reflective of reality will be the results. Don't believe that since you've "got" the user you can ask them every question you always wanted to ask them; your users realize that your survey is an expenditure of their time for a pretty small payoff (the slight chance that their data will effect some change in a product or service).
- It is certainly the case for me, and I suspect others as well, that the sheer number of feedback I'm asked for makes me engage in only the ones for which I feel very strongly (positively, or especially negatively), which skews the results.
- Don't say it should "only take about 5 minutes" if it has more questions than can even be *read* in that amount of time.
- Consider using binary choices ("I am [] satisfied [] unsatisfied") to gather data, rather than a 1-10 range or something similar. When you use a choice, you can report that "customer satisfaction is up 3% compared to last month", instead of saying "customer satisfaction rose from 72% to 78% from last month to this month, but the standard deviation is 18% and rose 7% from last month". Keep it simple.
- When I see a big grid of a dozen 1-10 range "how important is this to you" questions, it makes me mad… see point number 1
- Ranking lists of things in order of importance is also something I dislike, because it takes a long time, and I never seem to be happy with the choices I make overall
Survey pitfalls and suggestions
Surveys are everywhere these days, and a lot of them bug me. Here's some tips for those creating surveys:
