This week my inbox was hit over the head with at least five consumer outreach surveys.  And all of them stink.  You know what happens when your survey stinks?  Something even worse than nothing.  I get partway though, decide that I can’t take it anymore, and I just close the browser window without submitting, which means the company gets nothing, and I waste my time on responses that go straight to /dev/null, and we all lose.

I was recently tasked to create a survey.  I’ve never really done that, so I did a lot of reading on survey creation, and married that with my own experiences as a survey taker.  From this I derived the following set of immutable survey creation truths.  Steal them all, and survey your users, and improve your world.


Put some time into explaining exactly what you’re planning to do with this information.  Actually mean it.  Don’t tell me it’s going to “make the product better”.  Use concrete language.  Make me care.  Make me believe this is actually going to accomplish something.  Convince me that I’m about to make a difference.

Set Expectations 

Right up front, tell me how long this will take, what kind of information I need to have ready, and be honest about it.

Five Minutes

If your survey takes more than five minutes to complete, you’re going to need a pretty amazing reason for me to remain engaged that long.

If you’re way over this threshold, take a hard look at what you’re asking.  Is there real value in repeatedly asking barometer “how do you feel from 1-10” style questions.  Can you arrive at the same results by doing deeper response analysis and making inferences with less questions.  You probably can.  Don’t make the user literally answer every question you have about them.  Lift some heavier weights on the data analysis side and lighten up the end user’s workload.

Do another round.  Do a 2nd-phase phone screen, or build a focus group to interview.  Would you rather have 1,000 half-ass submissions, or 25 honest in-depth responses from users who opted-in to a second round?

Thin your Scales

Five, seven items tops, but I think even five is a practical limit.  If you are trying to learn something from ten and twelve point scales from very very very bad to very very very good, you’re going to learn that users will start blowing through answer grids just to put the thing out of its misery.  I honestly think you learn the most by analyzing the extremes of the scale, so loading up the middle with eight more choices is just chewing up everyone’s time.


If there’s one thing that turns my stomach every time is the never-ending flow of questions on a single page with no ability to see the finish line.  This is like your waiter walking up to you and just shoving the entire cheeseburger right into your mouth, hoping a) the whole thing will fit b) you’ll actually eat it c) you’ll come back to eat again.

Chunk the thing up!  Multi-page surveys are your friend.  They deliver questions in manageable bites.  They allow for progress meters.  They allow you to prevent users from seeing unnecessary questions based on previous responses, sparing them from having to check “N/A” on things you should already know are “N/A”.  And most importantly, they give you the opportunity to isolate a question that requires a thoughtful response, free from the distractions of your other questions.