Jan 5, 2010 6
A friend of mine just asked me for some advice on sending surveys. This is the list I came up with.
Sending surveys is an important part of early customer development; it helps you test a hypothesis and delivers you “perception” data. You can track how a user interacts with your service, it’s harder to track how they perceive it without surveys.
Early on in development of a consumer facing product, I’d recommend sending out simple surveys at short intervals (1-4 weeks) to a subset of your userbase. Below is the advice I gave my friend, if I’m missing anything, please leave it in the comments (Hattip to Hiten Shah, Leonard Speiser, and Sean Ellis who heavily influenced my thinking on this through tweets, posts, and conversations on this topic).
First thing is too check out this: http://venturehacks.com/articles/measure-fit (not shocking that Venture Hacks has the go to resource, is it?)
Then check out the tool that video is about: www.survey.io
But here’s my advice:
- Order your survey intentionally. Use early questions as eventual filters. If your second question is “how bad would you feel if you couldn’t use this product” that helps you sort later questions (i.e. my power users think this is the key feature, everyone else thinks it’s something else).
- When evaluating the data, you don’t want to optimize for the largest segment, you want to optimize for the segment that’s most engaged.
- Don’t ask any questions without understanding how you’d apply the data you’re collecting.
- Ask some open ended questions. The open-ended everything survey recommended in the post is a great way to start. But I like to have less than half my questions require typing — and it’s usually just an “anything else you think we should know?”. You get much higher response rates. But, there’s definitely a time and a place: open-ended questions are really useful for messaging questions and for early “discovery” surveys. Those questions also allow you to learn a lot more about the user (and how committed they are — you can tell a lot from the length and quality of their response).
- Ask for the ability to followup by phone, and do phone followups with every person who says yes. You’ll learn a lot in that conversation — and you’ll develop deep relationships with potential customers.
- On the subejct of developing relationships, provide an option for opting-in to the “elites” club — let them self select into beta testing groups. These elites can often become marketing assets. Yelp did a great job of this. David Barrett at Expensify is also doing this well right now. Survey’s aren’t just data, build a marketing asset.
- Ask for their reactions to the product, optional and freeform (limit text length if you would like to use in materials or on twitter). Ask “can we use this for marketing purposes?” Another idea: followup and get a small photo, name, link, etc. — use these assets to personalize the testimonials when you put them online.
- Don’t put explanatory text in front of questions. It’s tempting to try and put people in the right “frame” — it hurts you in the long run. Don’t alter the answers they want to give you.
- Short surveys win. <10 questions. <5 mins to complete. half that is much better. Don’t write an SAT test.