Information architecture (IA) is how content is organised online, including how information is grouped and labelled. In a well-designed website with good IA, information is presented under logical headings and pathways so users can easily find what they need.

You can find many online guides to developing good IA. Briefly, good IA and content should be built from 3 perspectives that are complementary and equally important:

  • Users. This perspective ensures that the website is easy for a user to navigate. It looks at who the main groups of users are, and at their
    • needs (i.e. what they want to know or what they want to do)
    • expectations and understanding (i.e. what they might already know about a subject)
    • behaviour (i.e. what pathways they are likely to use to find information; user behaviour is usually discovered through user testing).
  • Content. This perspective ensures that the website presents information in the most effective way. It looks at whether
    • the groups and labels will work well for the content (i.e. whether they reflect the site’s aims and messages, and the breadth, depth and meaning of the content)
    • there will be a good spread of content in each category (i.e. not 60 pages in one category and 2 pages in another)
    • labels have parallel construction
    • the structure is robust enough to allow content to be expanded without decreasing navigability (e.g. if a heading is too general, it may be added to over time and become a long list of subsections that is harder for users to navigate).
  • Context. This perspective is about the goals and constraints for the site, such as
    • what the site owner wants to say or achieve with the site
    • whether there are any rules about content, such as accessibility guidelines
    • whether the site will be regularly updated or expanded.

IA structure options

Depending on the subject of your content, there are many options for how to structure content.

Online structures a user may expect or be familiar with include the following:

  • Sequential. This structure takes the user from start to finish, in a step-by-step order. This choice makes sense when there is a set reading path, such as steps in a process.
  • Hierarchical. This branching structure is the way most websites are designed. Broad categories are presented up front, with an option to ‘drill down’ through the content. Pages are interlinked and often cross-referenced, but there is a set hierarchy of levels.
  • Matrix. This structure creates many links between elements of the material, with multiple possible paths to various parts. Matrix structures can facilitate browsing experiences. For example, one user might start with the oldest material and move to newer content, while another might access basic-level information before reading more advanced content.

Categories

Categories are the headings placed on content pages. Within each category, the topics should form a logical cluster and meet the user’s expectations of the links that would naturally fit under each heading. In general, IA categories can be:

  • topic based – these are usually nouns that break the main topic of the website into subsections (e.g. services, projects, news)
  • task based – these are usually verbs that guide users to pathways about particular activities (assessing your home, designing for sustainability)
  • audience based – these labels usually reflect groups of users (e.g. consumers, healthcare professionals, researchers).

You do not have to choose only 1 of these approaches – for example, you can have both topic-based and audience-based labels in your navigation. If the website is complex, you can use 2 sets of navigation lists (e.g. one at the top of the page and one in the main menu bar).

You can also use different structures at different levels – for example, the main site categories may be topics, but some webpages within the topics may have sublevels listed alphabetically or by region.

Shallow and deep structures

Content structures can be shallow or deep:

  • Shallow – many categories with fewer levels. The ‘3-click rule’ suggests that no content should be more than 3 clicks away. Users will usually find a reasonably shallow structure easier to use, although this depends on the content.
  • Deep – fewer categories that go down through many levels. A deep structure can work for sites with extensive content. However, users may be sent down ‘rabbit holes’ as they click down through many levels, and these can be difficult to navigate out of. To manage larger websites, it can be useful to think of other navigation aids such as multiple menus or tagging, rather than relying on deep structure (see Navigation for online content).

Shallow

Community
     Entertainment, concerts and fairs
     Library

Outdoors
     Parks
     Bushland

For residents
     Building permits
     Waste collection
     Pets

For businesses
     Liquor licences
     Applications to modify a structure
     Waste collection
     Council support for local businesses

 

Deep

Residents
     Community
         Outdoors
             Parks
             Bushland
         Library
         Entertainment
             Street fairs
             Concerts
     Council services
         Permissions
             Building permits
             Registering your pet
         Waste collection

Businesses
     Permissions
         Liquor licences
         Applications to modify a structure
     Services
         Waste collection
         Council support for local businesses

Drafting IA

To draft an IA, you will need to think about the 3 perspectives (users, content, context). Good IA is usually achieved through an iterative process of drafting, testing, updating and retesting, taking these perspectives into account.

The first step of drafting IA is sorting the content into logical category groups. There are various methods to achieve this, including:

  • just thinking about your content, writing likely category headings and sorting your table of contents into a logical structure
  • open card sorting – put example content on a set of cards and ask users to sort the cards into groups of similar content; label the groups once users have finished sorting
  • closed card sorting – develop a set of categories and put example content on a set of cards; ask users to sort the cards into the predetermined categories.

When you are thinking of categories, put yourself in your users’ shoes. Every user has a different perspective and different way of looking at the world. They are also likely to know less about the topic than the content authors.

Categories are not ‘right’ or ‘wrong’, but rather depend on the needs and understanding of your users. Any category that helps the user to find what they need is ‘right’.

Some useful tips when drafting categories are to:

  • use general definitions and groupings, not technical (e.g. tomatoes go in ‘Salads’, not ‘Berries’)
  • use common terminology, not technical (e.g. ‘Rain’, not ‘Precipitation’)
  • use concrete terminology, not abstract (e.g. ‘Patent applications’, not ‘Innovation’)
  • use categories that are specific enough not to become a dumping ground for content (e.g. ‘Research projects’, ‘Training programs’ and ‘Community outreach’, not ‘Our work’).

Testing IA

It is important to test and adjust your IA to make sure you are meeting the needs of most of your users.

To test IA, you ask users where they would search for particular types of content in your content categories and subcategories. This testing can be done in the form of scenarios (e.g. You are a parent trying to find out if the HPV vaccine is safe – where do you look for that information?; You are trying to find the bus timetable for Wollongong – what would you select?).

If users have trouble finding content, this is a sign you need to change your IA. You can think about changing:

  • the structure – maybe the grouping is not what users expect
  • the heading – maybe the label is not clear enough or the wording is not what users would commonly use.

Keep testing and updating until most users can find what they need.

Also see User research and testing.