There is no “right” number of questions to ask in a Web survey, but here are a few guiding principles.
You can include a lot of easy questions
It takes very little thought to answer questions about your gender, age, and country of residence. It may take a second more to consider whether you are willing to share your total household income. And you know really fast if you have bought a particular product or have called their tech support in the past week.
Just keep the options simple. For example, listing every possible country in the world in a drop-down list is complicated so consider asking about regions instead if the project goals allow. It also helps to put these kind of easy questions on a single page so there isn’t time spent waiting for pages to load.
Use rating and ranking questions with caution
Markettools did a great webinar a while back. (I can’t find the link unfortunately, if someone recognizes this study please send to me and I’ll add the link). They analyzed the behavior of thousands of survey takers to see where drop-off happened. The glaringly obvious finding was that rating and ranking questions (those huge matrices with lots and lots of boxes to check) make survey takers go away.
Obviously there are times when research goals demand a rating/ranking approach, but do your best to limit your use of those types of questions when another approach will work. A checkbox series may be simpler to complete even though it is actually more questions. Minimizing the number of options in a rating/ranking question also helps.
Consider the motivation of your audience
Customers who are giving feedback on a product they use on a regular basis will have a lot of tolerance for a large number of questions and in fact may welcome the opportunity to answer many questions that will simplify their use of your product in the long term.
We’ve done very successful customer surveys that offered no reward and took 20-30 minutes to complete. The audience was vested in the topic and was engaged. On the other hand, we often do Web surveys to gauge attitudes of IT professionals. In this case we offer a copy of the final report. We keep these surveys to less than 20 questions with at most one rating/ranking question and one open-ended question. It takes about 3-5 minutes for a typical IT professional to do one of those kinds of surveys, which seems to match their motivation level to get the report.
The Metric that Matters: Time to complete
If your metric for building surveys is number of questions, you can easily end up with a complicated survey that will give terrible results because it’s confusing for participants and impossible to analyze. Instead track the time to complete the survey.
Get somebody to take the survey who has not been involved in the development but knows the topic and see how long it takes them. You might be surprised to learn that your “3 minute survey” actually takes 15 minutes. Much better to find out before you field rather than later, when nobody completes the survey, or you get comments about survey length instead of value-added feedback.