Pros and Cons of Personalisation

If there is one subject guaranteed to get two web designers arguing, it is almost certainly personalisation. The promise is obvious - a website tailored to each individual who uses it, highlighting items that will be of interest to his or her particular profile, and consequently saving the users time and providing a superior user experience.

Alternatively, why not let the user tinker with the interface itself, bringing commonly used functions to the fore and even changing look and feel until satisfied with the result. Working on the principle that the user knows best, isnt this the obvious way in which to maximise the usability of a site or application?

Unfortunately (and rather inevitably) things are not so simple. Its probably fair to say that in an ideal world, perfect personalisation would be of great assistance to most users. But unfortunately, in the current web environment, too many things stand in the way and make personalisation a risky business:

Whilst users may be vocal in demanding what they want, they may have more difficulty identifying what they actually need. This can cause particular problems when applications or sites enable users to manipulate the interface directly. The end result is likely to be a mish-mash of desired features, with little thought to the overall structure of the interface and the relative prominence of interface elements.

Services which automate personalisation are hampered by one central issue - a lack of sufficient information. For example, e-commerce applications tend to rely on purchasing history. But this is a singularly poor indicator of what a user may be interested in next. Indeed, on reflection, the very fact that an individual has purchased a certain item may mean a similar purchase is less likely in the future. My own experience (with Amazons recommendations engine) suggests that due to a limited data set, these services are rather unimaginative and are only able to tell me what I already know.

Online publishing solutions which enable users to select those areas in which they are most interested can provide some benefit, particularly when they enable certain information (specific stock prices for example). However, a successful implementation of this type of system requires a watertight classification system that can guarantee to the user that he or she will not miss stories of interest as a result of decisions taken when personalising the service. As most readers tend to (at the very least) scan each section of most publications, it can be mistaken to remove them entirely - even if users believe this is what they want.

Applications (such as Microsoft Office), which automate personalisation by prioritising commonly used functions can improve efficiency. However, many users will find that interfaces which are effectively changing over time may be difficult to learn - and more importantly undo any learning that has already taken place.

Ultimately it is the designers job to understand user requirements and build an interface that enables any individual to find what they are looking for as easily as possible. Although personalisation is often a tempting course of action, it should only be considered in certain circumstances. It is too easy to fall into the trap of using personalisation to force hard decisions onto the user themselves.