2004

Websites are put to the test every time someone visits them, but the software used to record the popularity of a government site does not assess how easy it is to use.
Sally Whittle explains why.

This article appeared in The Guardian, Wednesday February 25, 2004 and The Guardian online e-Public section.

The chief executive wants to know how the website is doing. So you ask the IT department to pull up the server records. They show that hits are growing, page impressions are up, and the search engine has been getting lots of use. Great, right? Before you pass those numbers on, think again.

The search engine's popularity could be a sign that users can't find the information they're looking for. High page impressions may suggest they're lost, testing link after link. You don't know because your organisation, like many others in the public sector, never asked people whether your website is easy to use. "The website you have is probably a miserable failure," says Jakob Nielsen, principal of the usability consulting firm, Nielsen Norman Group. Nielsen has spent the past decade crusading against complicated menus, badly designed pages and poorly worded home pages. He says that 80% of public sector websites fail the usability test, meaning that the average user can't find and use information quickly and easily.

Usability is a particular issue in the public sector, in which websites need to be accessed by the whole community: the elderly, people with low literacy levels and those with visual or other disabilities. Managers in the public sector have a responsibility to ensure that everyone in the community can access online services, says Mark Gardner, website manager at Birmingham city council. "The site is paid for by its users and we have to be able to justify their investment. That means usability has to take priority over design."

When Birmingham redesigned its website two years ago, improving usability was a top priority. The council worked with a specialist market researcher to conduct focus groups, asking the community what it thought of the new site. "We got a lot of feedback, particularly on the menu and the terminology we used," says Gardner. "We used that to create a new menu based on eight categories of content that people can easily understand."

The focus group is an essential component of user-centred design, which is the holy grail of website usability. "The idea is to build your website around what users want and how they think, rather than how your council is structured," explains Catriona Campbell, managing director of The Usability Company. "When I visit the council's website, I don't know which council department I want to contact. I just want to know when the rubbish is collected on my road."

The first step in improving website usability is to think about how information is put together on the site - something the experts call information architecture. "Does one page follow another in a logical order," asks Campbell. "When I click on a link, do I have an idea of where I'm going to end up?" One way organisations can improve information architecture is to write the headings of web pages (or groups of pages) onto pieces of paper, and ask users to organise them into areas they think are related. "This is a really simple and remarkably effective way to devise a menu structure," Campbell says. Where users don't agree on where a page goes, information can be duplicated or a link created between the two sections.

Once you have a basic structure in place, Campbell advises creating a paper prototype of a typical web page, and again testing the site's design with users. Once this is done, create a prototype website using basic hypertext mark-up language. It might seem like a lot of hassle, but Campbell points out that early user testing often works out cheaper in the long run.

"It's a lot cheaper to move a search button on a paper prototype than it is to move a button on a fully-functioning website," she says. Early and frequent testing can offer warnings on potential problems. When Medway council redesigned its website last year, the three months spent on user testing helped reduce overall development time by half, according to website manager Sean Hale. "It was absolutely worth it because it opened our eyes to issues that we would never have considered ourselves, because we know the site and the information inside out."

The council initially videotaped a series of users trying to accomplish specific tasks on the existing website, and used the results to create a priority list for the new site. The usability team then returned to the users at every stage of development, while a specific disability working group considered how well the site worked for people relying on text-to-speech web browsers and other assisted technology. However, user testing can present managers with problems. "You'll always get people who say everything is rubbish.

People contradict themselves all the time and nothing is ever going to please everyone," says Hale. The best advice is to take the general point of view, unless an individual points out an issue that you are obliged to take note of, such as access for people with specific disabilities. If your website is up and running, it's still possible to dramatically improve usability with a few inexpensive and simple changes. For example, traffic monitoring software can reveal what pages on a website are being visited, and where visitors leave a site. This can reveal services that aren't being used or particular stages in a process where users become confused or lost.

The National Endowment for Science, Technology and the Arts (Nesta) is responsible for awarding grants to innovative community projects from National Lottery funds. However, traffic analysis showed that visitors were often leaving the site without viewing information on how to apply for grants. "We didn't know we'd hidden it, but it was apparent people couldn't find the information," says Maria Stark, Nesta's web manager. Simply moving this information further up a menu structure, or providing a link from the home page is often enough to improve usability. Traffic analysis tools are widely used, but it requires a degree of skill to interpret these results, says Nielsen. A page might be unused because it is hidden, but it's also possible that people don't need that service.

Alternatively, visitors may end up misinformed by content they have found that is poorly written. In the end, Nielsen says, user group testing is the only way to really be sure your website passes the usability test. "There's a tendency to prefer nice, clinical data to the messy process of sitting down with real users," he says. "But remember: they don't smell, and they won't bite. Who knows? They might even be nice about your website."

Return to newsletter