Create information clusters, not pigeon holes

Creating an intranet structure is both an art and a science. For best results it is important to follow a robust design process from beginning to end, meeting the needs of all stakeholder groups: the end users, the design team, the authoring community and senior business owners. A good site design is best done from the bottom up and though the resultant information clusters might look like pigeon holes, thinking in this way during the design phase can see the IA unravel.

Build from sound foundations

The best information structures come from taking a bottom-up approach. Information for staff should be clustered in ways that make sense for them. Best practice methodologies are based on this approach and use techniques to:

• determine the key needs of staff in advance by doing research
• understand how staff think about this information and cluster it accordingly using card sorting, natural categories, operational insight and other organisational factors
• structure the information in this manner and then test with staff to see if they can complete representative tasks, using tree testing
• design the pages to house the information and then test again when visual styling is completed, with usability testing

At several stages during this process, particularly after card-sorting, clusters emerge and are labelled. These now look very much like pigeon holes and the danger is that they are treated as such by ‘encouraging’ new information to fit into these groupings.

Labels can mislead

But the label for the cluster, particularly at the preliminary stages, is just a representation of what the cluster contains, and is often only one or two words in length. It is what the detailed items mean to staff that really defines what the cluster is all about. Once the cluster has been labelled, it is all too easy to just rely on the monicker to guide ongoing decisions. This is a natural thought process, but is based on heuristics, or cognitive short-cuts, avoiding the need to work with the detail. Losing sight of the detail makes it easier to misinterpret the cluster’s meaning, presenting the risk that the content of the cluster will begin to follow the title itself, rather than stay true to the intent of the original classification.

Here are some ways to avoid being too focused on the cluster label:

• when the clusters are being created, ask those involved to articulate what the cluster means, and what is should contain
• before deciding on the name, create a sentence to describe what it contains, then pick a shorter name to reflect this
• continue to regard the title as ‘flexible’, even if the boundaries of what it represents are fixed
• when adding items, give consideration to the entire contents of the cluster
• avoid the temptation to change the name ‘to fit in’ additional items
• if you need to change the name, do it based on a holistic view of all the contents, not based on the addition of one or two
• take a broad perspective across the full IA, and ensure that chosen names do not conflict or blur with other areas (also watch for gaps)

Avoid last minute meddling

Changes can be particularly damaging towards the end of the design process. Often a wider group of individuals become involved and want to influence the structure and terminology, but are not familiar with all of the work and decisions that have led to this point. This can be a particularly thorny issue if the influencers are senior, and have significant authority.

If this is a likely risk, then involve these influencers early. Canvass for their input, familiarise them with the process, and be sensitive to their needs and expectations early.

The design process by definition involves a degree of tuning and moulding, so working through these issues during development is normal. Being familiar with the issues mentioned above will also be of value if changes are required after implementation.