Online surveys
November 27, 2001

I've been working on a couple online surveys lately. I've been doing some research into how best to present online surveys. Here are some guidelines, tips and resources that I have been able to in the course of researching, executing and analysing the responses.

A good place to start:

WHAT THE GURU SEZ

Here's Jakob's 2 cents.

FREE STUFF

There are numerous sites out there that offer FREE online survey tools. Here are two that seem to be pretty popular.

SurveyMonkey
Zoomerang

Both offer fully customizable forms, and useful result pages. SurveyMonkey seems to have slight edge over Zoomerang in that they offer the options to distribute the survey over a couple of pages, so that you can even have one question per page which may be useful. However the pages don't show which or how many pages there are, which may lead to users bailing mid-way through the survey.

Yes, they both have absolutely awful graphics.

SOME GUIDELINES

Here's a couple of things I learned from developing online surveys for the two projects:

1. Give users a heads up: Tell them that they will be seeing a survey on the site. Ask for their participation. If you can, provide incentives for participation.
2. Make it short: Users take about 1/10 second determining whether a survey is worth their while. Long surveys are a no-brainer. Click - close window.
3. Don't make the users think: Ask questions that users will have immediate responses to. Word it carefully so that they can immediately respond - without thinking about what the question means and how they should answer it.
4. Avoid blurry multiple choice questions: Questions like: "how useful is this site to you?" - Answer: a. Very, b. Somewhat, c. Not at all - especially in a long list. I for one will click "Somewhat" all the time because I don't want to be too harsh, but I still want some change on the site. Sometime "yes/no" questions are more effective in situations like this, or make the questions more specific.
5. Similarly, if you are going to use multiple choice questions, try not to provide a "middle ground." Get the user to state whether they feel positive or negative about the question, with choices that are: "very satisfied; somewhat satisfied; somewhat dissatisfied; very dissatisfied."
6. Take results with a grain of salt: In general, if you get a 50/50 negative to positive response from users, this spells good news! Why? Because frustrated users are more likely to respond to feedback. Of course it hasn't been proven but for every satisfied user there are 2 - 3 dissatisfied users. This doesn't mean forget about the complaints, just don't think your site a failure because you are getting equal amount of complaints per praise.
7. Following up on point 6 - there will be those who will purposefully provide erroneous information or provide information from what they think the answer should be rather than what they actually do.

Survey responses are not absolute, but provide a good indication as to user expectations and behavior.

POP UP vs. EMAIL vs. WEBPAGE SURVEYS

We've found that in a small dedicated site, popup surveys are quite effective, despite all web usability gurus denouncing them left and right. Here are some popup survey guidelines:

- As mentioned above, provide users with a heads up. Mention upcoming popup survey in email newsletter or announcement section.
- Have a static survey page as backup just in case there are user who have closed the survey by accident.
- Make it short. About 8 questions max.
- Use Javascript/cookies to make survey pop up only once per user.

Email surveys (surveys included in body of an email) are, in my experience (asking student for 9 semesters to fill them out at the beginning and end of course) not very effective. I tend to think that it's better to make surveys links to webpages in emails wrapped up in a little candy (incentives).

Webpage surveys (surveys in the form of one or multiple webpages), all the rules mentioned above apply. Especially, keep surveys short. If you have a captive audience (that have been "encouraged" by higher ups), surveys can be longer, but still take the initial list of questions and cut it in half. Then make the survey one page. There is a certain gratification in completing a survey that's one page. If it is going to be multiple pages let the users know before hand how many pages and questions there will be and show which page they are on (e.g. page 1 of 3.)