I have a modal used to add a responsibility profile to an element. The responsibility profile is one item from a potentially huge hierarchical list with possibly thousands of items not listed alphabetically, but hierarchically. This makes a standard select list unusable.

I have two options:

  • a series of dropdowns where the selection from on dropdown populates the next (option A)
  • a select list with search functionality (option B)

To make matters worse the level of user skill is wide. Whereas Option B might work best for the advanced user, Option A might work best for the novice user.

Can anyone give me an example of how this dual approach has been used? Or another idea altogether?


download bmml source – Wireframes created with Balsamiq Mockups

  • 1
    One suggestion would be to combine them: Use a searchable/filterable tree view.
    – DA01
    Aug 13, 2013 at 21:52
  • Thanks for your advise. My original instinct was to use a searchable tree view, unfortunately this tree could be dozens of levels deep. But I may revisit that thought.
    – rdellara
    Aug 13, 2013 at 22:00
  • @DA01: You should post this as answer; it's worth an upvote in my opinion. Aug 15, 2013 at 15:06

2 Answers 2


I think of something like this:
enter image description here

Advanced users could search for an item while novice users walk through the hierarchy elements in a wizard-like way. Each next step depends on previous. At least, displaying many items brings more observation and searchability.

  • Thanks, this is an interesting idea. Can you elaborate on the use of the breadcrumbs? Is this similar to my option A with a search added?
    – rdellara
    Aug 14, 2013 at 17:44
  • 1
    Breadcrumbs reflect the way user passed through hierarchy levels before getting the current view. So user can watch the hierarchy branch from root to leaf. Yes, in this sense it's similar to A option. Aug 14, 2013 at 19:15
  • I get it. I like the idea, I think it won't work for my current purposes because the breadcrumb items could potentially be quite long.
    – rdellara
    Aug 14, 2013 at 19:41

The user-friendly option is search. Sadly, the serch has to be good to be of use.

When a user wants an artefact, they are happy to just get it. Creating one's own good storage concept and applying it consistently is a hard cognitive task. Understanding a storage concept created by somebody else is even harder, and that is what you are doing in Option A.

There are countless examples for this, in and outside of software. Imagine going to a library in search of a specific book. Is it easier for you to start looking for it based on your knowledge of the (very logical and well thought out) ordering system, or do you go to a librarian or a catalog machine to find out its exact location? Even if you go there for browsing, you probably don't want to browse everything, and having somebody point you to shelves with the type of book you are interested in is easier than holding in your head that the ancient Greek drama is in the back left corner and gardening advice is in the middle left shelf in the second room. Even as an expert user, getting an answer to a concrete search query is preferable to navigating, because navigating is easier for you than for a novice, but not as easy as just being told what you want without the need to reconstruct the information from an abstract navigation schema.

Some prominent examples in the software field include:

  • Web search. Option A was widespread in the early days of WWW, with Yahoo becoming one of the early dotcom stars based on this principle of categorizing the Web. As soon as a convenient, simple search (Google) came along, their popularity started rapidly declining, even though they switched to the search model later. But as you will notice, a search had to offer a certain minimal level of usefulness before it got accepted, others tried this model before PageRank and failed.

  • Launchers on desktop computers. The start menu has been industry standard since Windows 95. After a few years in which the Mac users had launchers, and Linux copied them, offering isle solutions more or less integrated with the desktop environment, Windows also caught up with the trend, offering a usable launcher within the start menu with Windows 7 and now even daring to do away with the whole menu for Windows 8. (I know that users hate this; this is just resistance to change. I have personally observed people who used to cry for their start menu suddenly realizing how much better the new way is). Again, this feature needed a minimal quality before becoming a viable alternative to the start menu. There has been a "run" box in Windows for ages, but having to type "Program%20files\Paint.NET.exe" is worse than a menu, while typing "paint" and getting the choice between MS Paint, Paint.NET and "PaintShadesGrandmasHouse-Renovation2011.docx" is great.

  • Amazon is an interesting example, because it offers you both. It has a good search, and also a navigation which is much more sophisticated than a simple tree structure, allowing the same book to be filed under multiple categories. I don't know about you, but when I need a specific book, I type the name in the search box. If I need a book on a topic, I search for a keyword from this topic, and sometimes use the recommendation of similar items. I have never felt the need to directly navigate to a category. I just tried it for this answer, and my first two tries failed, even though I tried looking for books from fields I am very familiar with. The problem, as anybody who has ever tried to box information into a hierarchical structure can tell you: The criteria I use in my head to divide books into categories have very little in common with the criteria the Amazon people used. Guessing the content of a given category was hard, and I ended up being wrong frequently. When I read "food cultures", I imagined books on e.g. Victorian English food. It turned out to be books on eating vegan, or wheat-free, etc. At the same time, when I opened a category which was obviously correct for the book I was searching, about half of the sub-categories looked like equally likely candidates to contain it.

So, if you want to offer your users a good experience, give them a search. But it has to be a good one. And writing a good search is terribly hard. Simple pattern-based search (which does not find the item "mouse" when the search query is "mice") is not good enough. If you are writing a desktop or web application, use a third party search library. I don't know if there are any available for a mobile applications. If you can't use a good search, you may have to test the usability of the navigation option against the accuracy of whatever search you are capable of creating within the project, and choose the option which frustrates your users less. But none of them is especially user friendly, and both will lead to a significant percentage of failed tasks.

  • Thanks for the thorough explanation. I personally use search for everything. However the types of content these users will be searching for is very technical (think codes) so a posible search might be 1002-TE. This requires a great deal of knowledge to find your desired item using search, which advanced users will have. Novice users might only know where within the hierarchy they are looking. I think this solution is going to require some combination.
    – rdellara
    Aug 14, 2013 at 17:51

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.