1) Validate assumptions . A survey shouldn't be a fishing expedition. Focus your effort by having some expectations in mind and then using the survey to test your assumptions. "Our rule of thumb is that research should validate 60% to 70% of intuition," Broetzmann said. "The rest should be new insight."
2) Match the survey type to your objectives . Online surveys aren't good for every purpose. Use Web surveys to get a general sense of what your customers or prospects think about a topic. Then drill down with more focused telephone research, which allows for deeper discussion. Use focus groups only when you want to dive deeply into a topic. "Many companies make the mistake of using focus group research first," Broetzmann said. "They're best for getting rich information. There's a reason they're called focus groups."
3) Make the survey interesting . A lot of lip service is given to survey length, but even long surveys can yield very good response rates if the questions are interesting. Asking people about a specific interaction they've had with your company, or stating that their opinions will be used to create or improve products, increases the chances of engaging the respondent. Most important, seek an audience that cares deeply about the survey topic. Broetzmann said even long surveys can yield a 50% response if the audience is engaged. "Think like the customer," he added.
4) Keep it simple . Avoid open-ended questions requiring complicated responses. People don't want to type long answers. Complex questions with a large matrix of responses or a long list of products will also limit results. Keep questions short and specific, and avoid complex, hypothetical questions that require people to make a lot of assumptions.
5) Avoid the most common questionnaire errors . One common mistake is asking demographic questions at the beginning of the survey. "It basically tells the person you're trying to sell them something," Broetzmann said.
Another no-no is leading the respondent with questions that mention your product name or use your marketing language. A third is to offer response options that are unbalanced. "If the choices are `excellent, very good, good, fair and poor,' you're basically offering four out of five options that aren't so bad," he said.
Also, avoid questions that ask for a rating on a one-to-10 scale. Five- or seven-point scales are more statistically valid.
Most online survey providers offer sample surveys or tutorials on questionnaire design. Doing upfront research will save you from making mistakes that could lead to dangerously misleading results.