The Tendo View

Putting a little thought and planning into your Web surveys will not only yield better results, but it can reduce the annoyance factor for visitors and offer solid information for improving both your website and your direct customer communications.

First Things First

Whether you conduct your survey via phone, snail mail, e-mail, or the Web, experts actually offer the same basic advice:

• Know your aims—It may seem obvious, but you should know what you want to learn from a survey before you start. The more narrow your aims, the more achievable they become.
• Learn just a little about statistics—Even if you work with an expert, learning something about sample size and how data is tested will go a long way toward creating a valid survey that achieves your aims.
• Learn about questionnaire design—You can avoid some common pitfalls (and a lot of frustration) if you learn a little about the basics of questionnaire design, such as the strengths and weaknesses of different questions.

Writing the Questionnaire

With an idea of your survey aim and a grasp on the statistics and the basics of questionnaire design, you’re ready to get started.

1. Keep it short—Typically, a visitor survey has two basic aims: Demographics (who visits the site) and desires (what visitors want), and nearly every question is a variation on these themes. Refining questions to eliminate overlap and non-essential information will keep the questionnaire short, and thus more likely to be completed.
2. Ask one question at a time—Make sure your questions ask only one thing, or it becomes difficult to discern which part of the “double question” your respondents answer. For example, “Please rate the timeliness and accuracy of what you read on our site” might lead some people to evaluate timeliness and others accuracy.
3. Run a pilot—Test your questionnaire on a small sample first. It will not only help you fine-tune questions, but can help correct other problems such as overly long questions that may cause people to abandon the survey. If you can’t run a pilot on visitors, at least ask colleagues who haven’t been involved in the design to run through it.

Executing the Survey

1. Ask on exit—Too many surveys make the mistake of hitting up visitors immediately. Besides being annoying—remember, your visitors came to do something else, not take a survey—it’s counterproductive. To keep your survey from interfering with the main function of your website, present it once business is finished.
2. Say why—As Web surveys proliferate, the percentage of people who actually say “yes” grows smaller. There’s no surefire way to counter this trend, but it helps to explain the purpose of the survey and how users will benefit from the results. As with anything on the Web, keep the explanation short, and if you can, let some personality shine through. If your voice and tone are irreverent or playful, your request should be, too.
3. Say how much—Visitors are also more likely to respond if they know how much time the survey will take. Offering the number of questions also helps. A word of warning, however: don’t cheat. If you have 10 questions but three of them are monstrous, multipart queries, you’re likely to get high abandonment rates.
4. Use a “progress bar”—A simple bar that fills in as the survey is completed can help keep down the number of questionnaire abandonments. Watching the bar fill not only offers the satisfaction of completing a task, but also helps visitors gauge how much is left.
5. Respect “no”—Try to keep from asking a visitor to take a survey more than once. Setting a cookie is a fairly simple way to make sure you aren’t bugging people who’ve already said “no.”

Improve and Repeat

If you intend to repeat your survey to keep in touch with visitors and detect trends over time, your next step is to improve it.

High nonresponse rates (the number of people who refuse the survey) are typical on the Web. However, high abandonment rates (starting the survey but not completing it) indicate that you need to rethink some things. Be sure to revise and retest your survey before repeating it. There are several things to look for:

• If respondents are exiting at the same point in every survey, you have a problem question. If they fall off more randomly, you need to reexamine the overall approach.
• Are questions too long? Are there too many on a single page? Are you asking only for the information you need to achieve your stated aims? If your goal is to improve usability, visitors may balk at being asked how much they spent on electronics last year. If the question is legitimate, help them understand how it relates to your aims.
• Repeating your survey on a regular basis helps uncover changing visitor profiles and needs. Keeping some questions the same will make one survey’s answers comparable to the previous survey, allowing you to determine if a change between one survey and the next is statistically significant, and offer the insight you need to stay relevant and engaging.