Information architecture performs better in real life than in testing

June 23, 2016

I was talking to one of our clients the other day who said that he was seeing a 94% task completion success on his site – needless to say, he was delighted! This was better than I recalled from testing, so I thought I’d dig back into the reports to see what was happening.

Before embarking on site redevelopment we had tested their old site for task completion and the success rate was only 45%. We then worked with the web team to redevelop the information architecture and tested the task completion rate again using the same ‘basket’ of tasks; this time the overall success rate had significantly improved to 75%, with the success rate for top tasks at 82%. Subsequently, when the redeveloped site had gone live, in a survey of about 1,000 people who had used the website 94% said they found what they were looking for.

When we’re developing a new information architecture we always test it with ‘real’ users to make sure that it works – we don’t rely on expert or internal opinions. We set a pile of realistic tasks for users to attempt, including tasks to test quite obscure parts of the site as well as links to the ‘top tasks’ – the content that most people are going to want to find most of the time. In reality, the top tasks are the ones that most users are going to attempt, and most of the rest of the site will only be viewed by a much lower volume of visitors. The overall testing task success rate averages all the tasks and is not weighted to the top tasks, so consequently can understate the success rate that can be achieved in reality – as is nicely demonstrated in our client’s example!