Recommended Survey Length
Posted by Jeffrey Henning on Mon, Mar 09, 2009
Last week, to my surprise, I was disappointed by a survey I was invited to take, because it was too short! It asked me only three questions: one clarifying my usage, one asking me how likely I am to recommend it, and an open-ended question to explain my rating. That was it. It reminded me of my brother-in-law, who will call me from work occasionally to catch up, then cut it short to say he has to go back to work (as if I had called him!).
In the case of this survey, the software company initiated the request for feedback, then "hung up" before I had a chance to really provide much input. They had my attention, and they squandered it, and ticked me off in the process. I had clicked on the link in the invite to take the survey because it was about the personal finance software I've used for years. While I'd admit it is much less fun to use this application than it used to be (as it resolutely tracks my holdings going down!), it is still an important program to me, one I use three to four times a month. I had quite a few points I wanted to make about where I thought they could improve.
The software is important to me, but the software company made me feel that my opinion wasn't important to them.
As an industry, we have heard the refrain from respondents that surveys are too long, and we've come up with ways to shorten questionnaires. I have joined the chorus of complaint, as far more often I see surveys that are much too long than surveys that are too short. But I think I need to adjust my message: surveys should be the right length. That three-question survey would have been the right length it had been about my oil change. And I have had customers happily answer 100-question surveys, which can be the right length as part of an annual account review, when the customer is spending $1 million a year or more with your organization.
After a couple quick searches, I couldn't find suggested questionnaire length by type of survey. Here are some ideas to start the discussion.
- 2-4 questions - Transactional Survey - A quick follow-up to a standard service interaction, such as a retail purchase, a customer service call or a billing inquiry, designed to provide ongoing measurement of service quality. Reasons to exceed this length: measuring conformance with a Service Level Agreement or analyzing behavior patterns to shape better experiences in the future.
- 5-10 questions - Event Evaluation - A more detailed follow-up than a transactional survey, acknowledging the respondent's investment of time and providing them an opportunity to rate speakers, venue and logistics. To be used to guide future event planning.
- 10-20 questions - Customer Satisfaction - An automated follow-up to a purchase, after the recipient has had enough time to use the product for a while. Provides the customer the opportunity to rate the product across the broad areas that comprise the overall product experience.
- 20-30 questions - Planning - An opportunity to gather detailed feedback of possible future direction for a product or service, to help the sponsoring organization prioritize.
- 50-70 questions - Major Account Review - An annual assessment of customer satisfaction with an organization's largest customers, often with multiple individuals across multiple departments. For really long surveys, consider using an executive interviewer conducting face-to-face or telephone surveys to gather the results.
- 70-90 questions - Employee Satisfaction Review - An annual or biennial measure of employee satisfaction, designed to prioritize HR initiatives and measure employee engagement.
As a unit of measure, of course, a question is as bad as a
league ("the distance a man can walk in an hour"): a question, after all, can be as short as a checkbox or as long as a multi-sided matrix. For practical purposes of following these guidelines, think of "a question" as the average single-select question. Treat that two-sided matrix with ten topics as 4 or 5 questions.
Now, if someone pressed you for the right length of a particular type of survey, what would you recommend?