Website Design Optimisation

[Stewart]

Perhaps this is a little early to throw into the pot but I decided to anyway. We're spending a lot of time talking about the initial design of a website and not much on what to do when its up and running to make it better! The days when slinging up a website and expecting users to flock are long gone and although sensible website design norms have developed so we know where to put key elements (or at least an have an idea where to start!) this doesn't guarantee user satisfaction, and certainly doesn't drive user advocacy or guarantee a successful business model.

We can all sit around and criticize the hell out of each others website, which is fun and useful and clearly has its place! But the wonderful thing about the web is that you can quickly measure the success or otherwise of design changes with the use of some very powerful and cheap tools. Again Google comes to the rescue with a powerful testing environment Google Website Optimiser - if you want to move an element on a page, leave out the image or change a call to action its possible to test the effectiveness of various design changes. You can test whether one page layout works better than another to achieve your goals (of course you need to define what these are up font!!) using A/B testing. If its at an element level and you have a number of change ideas you can test them altogether using multivariate testing, but of course a multi-element multivariate test is going to need more participants to gain a statistically significant result than a single element experiment.

When you move into this it's important to have a grasp of the maths but fortunately Google does it for us and highlights statistically significant results. Admittedly it difficult's to arrive at statistical significance if your unique visitor numbers are diddly squat! However I'm a firm believer in taking decisions where possible based on significant user defined probability rather than gut feel. By all means come-up with the ideas based on gut feeling but test, test, test rather than implemement at haste and regret at leisure!!!

If nothing else remember that testing is key to improving user experience and its easy to do by inserting a bit of javascript , loading the element(s) and letting Google do the leg work. Be interesting to hear of anyone else's testing experience?

[Dreya]

HI, very useful tip and actually very interesting points you have made!

[David]

Stewart, you make some good points. I have to say that given more time, we would cover this important new area of web business/design. In the meantime, there are some useful new books covering this stuff. Web Analytics 2.0 by Avinash Kaushik is particularly well though of although I've not read it myself.

[Stewart]

This is one of the best books on the topic that I've read. It also gives a great overview and commentary on the issues facing many a corporate website owner. I'd thoroughly recommend reading for anyone who wants to appreciate what keeps larger clients awake at night - useful insights if you are trying to sell them your services!

The issues are in fact the same for any website large or small and with free analytics services providing access to detailed web analytics data it is equally valuable to understand the topic as an individual website owner.

I wonder given its importance in the ongoing design process if analytics shouldn't form a significant part of a future content Mgt module?