Conducting a Survey (things to think about)

We’ve written many articles around conducting surveys and ways to increase response rates etc. This blog post is partly an answer to Ivana Taylor’s comment :

Have you done any posts on how to construct a good survey from scratch? I know you have lots of articles – but I’m specifically thinking about the tendency to ask too many questions then get garbage in – garbage out.

Surveys are an extremely broad topic. If you think about it, our customer base is so diverse that it is a challenge to design “best practices” without mentioning a bunch of caveats. For example, surveys are used for Employee Satisfaction Measurement (Exit Interviews etc.) in the Human Resource space as well as a Customer Satisfaction Measurement instrument. In one case, you can afford to have a long survey (employee surveys are almost universally getting response rates of 40% or higher even when they are 40/50 questions long) whereas customer surveys are usually in the 20-40% range on a great day with well defined incentives and a short survey.

That being said, we’ve put together some information to get you thinking about designing your surveys. Here are a couple of articles:

These articles should get you started on thinking about the design of online surveys.

Since many of our customers are into measuring “Customer Satisfaction” we’ll dive into that. We’ll talk about Employee models in another blog post.If you are running customer satisfaction surveys, I highly recommend that you take a look at the Net Promoter Scoring model (View QuestionPro implementation here). Although Net Promoter has its own set of detractors — the business of customer satisfaction measurement is indeed a high stakes game. Some consultants will argue for a complex model for calculating “loyalty” (for example the ACSI/Michigan State has it’s own “proprietary model”) and others yet argue for a simple model like the Net Promoter Score.

I think, part of the reason (apart from the simplicity) that Net Promoter has become really popular is because of the lack of other “published” models. I think Bain/Fredrick/HBR changed the game when they openly published the Net Promoter model unlike the ACSI which “licenses” its model.

We (QuestionPro) personally prefer to use the simple model (Would you recommend xxx to your friends colleagues?) and then keep a very close eye on issues coming out our support team to measure satisfaction.

One response to “Conducting a Survey (things to think about)

  1. Ivana Taylor

    That’s fantastic guys – thanks.