5 rules for creating a great survey

Most feedback surveys are badly constructed.

When some people realise that the main component of what I’m building is based on asking for feedback from website users, they respond with a sniggering I never reply to those things. Beside being a kind of a useless feedback to give me, it represents a very real problem that is taken way to lightly. Most feedback surveys are badly constructed.

The main reason why most surveys fail to provide any tangible value is that they’re simply not human friendly. It’s all too easy to forget that a feedback survey is a human to human conversation at heart. It’s actually the equivalent of asking your friend for a favour, yet that is often lost in the process of deciding on which questions to ask. Since the fact is that you are asking a whole bunch of people to do you a favour you are obliged to respect their time and intellect. Here are some rules that will help you a long way towards not doing a lousy job: Only ask for information you can’t find in other ways Only ask for information that you actually plan to use

  1. Only ask for information that is actually important, not just nice to know
  2. Always use the responses to improve your service for the users that kindly answered it
  3. Respect anyone that doesn’t want to participate; Stop asking them
  4. But what does this actually mean? Lets exemplify.

Only ask for information you can’t find in other ways means that if you have a logged in user with a profile where for example age is set, you don’t ask the users birth year even if you run the service through a third party service. Asking for information you already have is disrespectful, and being disrespectful will result in less people completing your survey, so both you and the user lose.

Only ask for information that you actually plan to use and that is actually important sounds obvious, yet we see all the time that it’s not the case. If you are asking your web magazines readers for feedback regarding the magazine layout, and you tack on a question like whats your sex or whats your annual income you are not being honest about your intentions, in the best case. It’s easy to see how this information is valuable to you, because you can resell that information to your advertisers! But its as disrespectful as you be towards your users.

Always use the responses to improve your service is all about taking action. Don’t run annual surveys just to produce some company internal slide deck about your demographics. If you’re not using that data to actually improve the service, then just wait until you have that as a reason. The world is full of constraints, and one soft constraint is how many surveys a year you can push on your users. Another is how many questions you can ask in a single survey. For your own good, if you max those constraints on useless for fun-surveys you’re just spilling the milk.

Respect anyone that doesn’t want to participate means that you need to accept that a lot of people are not interested – right now, or always – in answering your questions. Be discrete about your request, and when they turn you down, just accept it and let them continue as usual. It also implies that you should not use invading visual designs to ask people. Its better to ask for feedback after people have done what they came for, instead of throwing a big blocking modal in their face from the get-go. Thats the type of survey user experience that have caused the default response of I never reply to those things.

These rules are quite simple, yet they are often not used. If you respect the people you’re asking favours of, then the flip side is that you will also make better surveys that receives higher quality responses. Humans are strangely glad to help, and if you only respect their time and privacy they will be helpful over time. So the next time you’re thinking about running a survey to get some insights make sure you do that for some better reason; Like a redesign, to prioritise the next quarters work, or to figure out why your organisation even exist.