Tags interface for an desktop accounting application?

Question

The ancient question of tags vs categories will probably never grow old. Of course, in its general form the answer is "it depends". So here's a specific case:

• The application is an established accounting and ERP package.
• The codebase is 15 years old, it has a large userbase.
• It currently uses standard menus ("categories"), but these are overcrowded already. There are probably several hundred items there, I haven't counted. Each menu item opens a new form.
• There is already a quick-search feature which is similar to that of Windows 7 start menu - as you type, it filters out the menu items that contain the specific substring.
• The users are mostly not-tech-savvy accountants; changes in the UI are often fought against.

So I'm wondering if tags could help out here, and how could it be done. Or perhaps it is already good as it is? Surely, for someone who's worked with it for many years there is no problem, but the learning curve for new users can be pretty steep, especially since different accounting packages tend to give slightly different names to things (or organize them differently), and finding stuff can be challenging.

Answer 1

My immediate inclination is that the slightly fuzzy edged search or navigation facility provided by the tags concept is not likely to fit well in the context of the precise nature of the world of accounting or resource planning. The user model seems slightly at odds with the more unstructured tags concept which more suits a changing environment.

The good thing about categories is that typically the categories either do not change or changes very slowly over time (depending on context of course). Thus if I am regularly using categories I know where in a list to find a category I am interested in.

For an established application with a large user base, in this environment it strikes me that tags might be a conceptual change too far.

My additional concern though, is that you say there is currently several hundred items in the existing categories, and that leads me to think that actually the key to improving the experience of the user would be to provide other routes to choosing an item of interest.

Introduction of favourites and recently used items in addition to the existing quick search might be relevant - and appreciated - here. Favourites might be pre-populated with popular or most common items for the benefit of new users. Other accessors such as short-cut customisation (accelerators) also help the more expert users who will undoubtedly find repeated access to large category lists annoying, as they would scanning tags.

My instinct would be to want to see the package in use (by real users) and watch them using the categories system and learn what issues and repetitive actions users are really making.