I don't expect that there's a one-size-fits-all answer. I hope this isn't too subjective.

I'll use this StackExchange page as an example. There are links for Tags, Users, Badges, and Unanswered. At the bottom there are links for a tour and chat. Assume that there's some expense involved with maintaining each, and you can't add unlimited features to the site.

What sort of thresholds are set to determine whether a feature is used and accepted? For example, if 0.5% of users ever try the "tour" feature, do you keep it or remove it? What if it's 1% or 3%?

Do you even set a threshold indicating what use you expect to see for a feature? At what point do you drop it even if it has some users?


  • 1
    I do not think there's a one-size-fits-all answer. I think it should be user based. Ask your users about their needs. Discover the pain points they might have in your application and go from there. May 5, 2017 at 20:26
  • What are you looking to accomplish? Reduce clutter? Introduce a different feature in its place? Assist with navigation? Consolidate? If you were to strictly base your decision off percentage then StackExchange should probably remove their Legal link. Providing you with this example is the only time I have ever clicked that link in my 4 years of membership.
    – MonkeyZeus
    May 5, 2017 at 21:00
  • 1
    I'd like to make a case that if no one uses a feature then we should delete it to reduce clutter and maintenance. It seems intuitive to me but I'm looking for outside opinions. Some people misapply UX principles and say never get rid of anything ever. I realize that constant, unpredictable change can frustrate users, but we're not doing anyone any favors by hoarding features that no one uses at the expense of building new ones. May 5, 2017 at 21:38

2 Answers 2


This type of question is often asked when people try to use one measure to encapsulate or predict change. My advice is that metrics should not be used as a measure of a target, but as a means of monitoring change. That is not to say just because a metric value stays the same that there's no change, but that you should try to pick measures that can better reflect a change.

So back to your question, firstly you are trying to associate a quantitative value (in this case amount of use) with a qualitative characteristic (valuable). How you measure the amount of use is already a factor, because the amount of use has more than one dimension (e.g. each usage also has a duration associated with it). On top of that, it is not easy to qualify what someone might perceive as being valuable because there are also many different factors that contribute to it.

So when you are using a metric that has multiple factors to derive at an outcome that also has many different factors, it really sets you up for a very complex analysis to try and tease out the variables and factors involved. My suggestion is to take a simpler approach to break the usage down into different components and use that to predict or model a single factor (e.g. download count or average rating rather than being valuable). That way you will have a more useful and actionable result instead of complex analysis that might not mean anything significant.


I would suggest understanding your expected usage and compare that to the usage you're getting.

Using your example of a "tour" feature - look at the percent of new users who are engaging with that feature, instead of looking at a percent of users overall.

But even better - use qualitative methods for determining if a feature is important. There's a difference between removing a core site feature due to lack of use, and a privacy setting due to lack of use. For the former, a lack of use could indicate any number of problems or lack of value. For the latter, it's simple a low-volume setting that should probably be kept regardless of usage.

Ultimately, value isn't dictated by the % of users using your product, but rather whether or not it's providing the service it's meant to in a way that users feel is beneficial to them in some way.

  • We're not quite at that point. We don't have an expected usage. Does that make it sound like we're not sure why we build the features we build? May 5, 2017 at 21:33
  • I think that even if you're not at the point of understanding your usage, my point about usage not being dictated by %s still holds.
    – kristinalustig
    May 6, 2017 at 3:54
  • I agree. For example, does it generate Sales? Do visitors return there? If everyone checks it out once and never goes back that says something. I'm just trying to get us to crawl before we walk. At this point one could accurately say that we flat-out don't care whether features ever get used. Sort of like McDonalds in Moscow. May 6, 2017 at 11:39

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.