Can you experts please let me know some questions that can be asked in a survey?

We have just released an updated version of our data security (encryption) desktop application but the response is not as we expected it to be, so I just wanted to convey a survey from users who downloaded our product but uninstalled it or didn't went ahead and purchased it. Can you guys give me some suggestions as what exactly the questions should be? Should I ask them whether they found the new version complicated or is it user friendly or not. Any suggestion would be a great help to me.

  • is this one for SUMI?
    – colmcq
    Oct 20, 2011 at 20:21
  • Sorry, I didn't get your point. Oct 20, 2011 at 20:28

4 Answers 4


you can start with http://sumi.ucc.ie/ as a template, but maybe tailor it for your application nuances. SUMI is a pretty standard framework for this kind of thing, though it is flawed insofar as it has drawbacks as with all subjective scales: it focuses mainly on software; scales mostly addresses classical usability issues, smaller part is about affect; results are not highly informative for redesigners

you could also construct your own proprietary questionnaire

both these approaches, but more so the latter, require statistical analysis of the data, specifically correlations and statistical deviations around the metrics you use.

One of the easier - and I use that term with caution- methods of analysing the data is to determine the validity of the correlations and deviations from various data points you observe from the survey results. Your results will not fall onto obvious points but will form clusters around a central, true value. K-means clustering http://en.wikipedia.org/wiki/K-means_clustering is way of identifying most likely correlations and deviations inside clusters: put simply, k-means clustering looks at every data point in a cluster and by algorithmic means establishes which data point in that cluster has the lowest set of standard deviations compared to others in that cluster; this gives you a less subjective definition of which data point represents the consensus among the users interviewed.

caveat: this approach only for large sample sizes where you can establish a confidence interval of p< or = .95


You might have some issues that warrant more than just a survey.

The fact that the response isn't what you expected to be is a concern from a preparedness standpoint - was the product tested before the launch? What information did you gather there?

Surveys are great as a supplement to overall UX improvement, but they generally aren't reliable as far as volume goes. It is also very passive in terms of collecting data, so you will be waiting for the responses to come in while the UX remains a mess and nothing is being improved.

I highly suggest getting, at the bare minimum, a quick way to openly express a user's feelings about your product going. From a product standpoint though, I'd be working with testers immediately to try and identify what the issues are with your product.

  • Thanks for you answer. Yes the product was tested well on all 32 and 64-bit platforms before being launched and there are no known issues to be found. Also users who are using our product and contacted our support didn't complain about any such issues. So, I personally feel user is either not finding it user-friendly or there is something about user experience that we have missed out. Oct 20, 2011 at 20:40
  • @FasihUddin I have a hard time grasping what you are saying. If there was a large, sweeping issue, you should have definitely unearthed some clues in testing. So right now you have no leads as to what is causing the exodus?
    – Nic
    Oct 20, 2011 at 20:49
  • Yes exactly, no clues whatsoever.. application was made user friendly based on the customer feedback received for the previous versions, feature suggestions from users were also added. Do you think giving many features in a single application is a drawback? Oct 20, 2011 at 21:28
  • @FasihUddin lots of features is ok, as long as they're not severely different than your end user's expectation. Is it possible to reach out to your early adopters and just ask them some basic questions about the new version?
    – Nic
    Oct 20, 2011 at 21:32
  • Yes, this is exactly what I wanna try but I am a bit confused as to what exactly to ask them. Whether I should ask if they like the previous version more? How do we actually get to know what a user expects from us. Oct 20, 2011 at 21:45

I just came across a product that does this kind of thing. Check out www.sugarsync.com.

I'm using the software for a project and had to uninstall to re-install. I got a survey like what you are interested in.

In doing a survey, you need to start with a few well defined goals in mind. Then you can begin to ask questions that will provide insight into how to reach those goals. For example, you suspect that the software is not "user-friendly" so people are not adopting it, but that is not very specific. I've experimented with using the SMART goal system (http://topachievement.com/smart.html) to develop survey goals; its not perfect, but it does help to avoid non-specific, vague goals. So, a more specif goal might be "We want to improve the adoption rate by 10% by the end of the year".

You might ask questions like "What comparable products are you looking at?" or "What product did you use instead of ours?" Then look at those products to get some insight.

When you do have a list of questions, make sure you get an expert in the language to edit/proof the survey. This is, I think, the one thing that gets overlooked most. I take a lot of surveys because I write a lot of surveys. Bad grammar and spelling just looks bad.

Lastly, look at and analyze any open-ended text comments people provide. There are programs that do this, but it can also be done simply in Excel. A "score" like a mean, median or mode will tell you that there is an issue with a particular feature. The open-ended comments will tell you what to do about that issue.

  • Interesting, I will be going through it. What is more important for a desktop application, a good color rich skinned GUI OR lots of feature compiled in one application OR it should be more user friendly and easy to use? Oct 22, 2011 at 5:27
  • That's a question you should be asking the people who download your software.
    – eBeth
    Oct 24, 2011 at 3:13

I would say the user who is uninstalling your application isn't in a mood he is willing to answer to a lot of questions. He already wasted too much time downloading and trying it out.

-> You have to be short if you want to get any answer.

Thus, you can't choose a usability scale survey and anyway you would get an overview of your application but you won't have any specific information about what exactly the user tried to do and what are the real issues you need to correct.

It's going to be difficult to detect what are the issues with a few questions so you should make an open question about what did he want to do and what went wrong or what didn't he like. Start the survey explaining that his feedback is valuable to you and it's gonna help you improving your application.


Please take the time to help us improving our application.

Tell us what did you want to do and what didn't you like.

-------- Text Area --------------------------------------------------

4/6 lines to answer, keep it visually short, we don't want the user to think we expect him to write a lot.

-------- Text Area --------------------------------------------------

Thank you for your answer. (on send)

Should you ask for his email ? I don't think so, keep it anonymous and they won't feel bad giving negative feedback...


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.