Search with Design

Not all stats all created equal. And Time On Page is one of those.As a designer, I’m always looking for ways to improve my work and continually that has been directed at delivering quality designs that answer the audience’s questions and provide intuitive design for navigation, content and images. One way to examine your effort is to monitor time on page. But, agh, the TOP stat is so misleading. So for example, I redesign a section of a site to create an easier experience for the user offering more obvious navigation, easy to digest content and a simple tour-style click thru button on the same place on each page. Viola! Great design, great implementation, and yet, the client argues, nothing was gained because the overall time on the page decreased. Is this a valid argument, or could I be right, that a better design makes content more consumable, quicker?

I hate the answer, both are right. So lets look at when time on page counts and when its worthless.

It counts

Flash sites - most flash sites are designed as one page, housing all the content and media in one browser-sized block. The greater the time on this page is also seen as the greater the experience. You could however add tracking to items that could be clicked on to better dissect the audience.

Trailers / Tutorials / Flash Demos - Here time does matter. How far did the visitor get? It’s much like the shopping cart effect. Watch to see where they dropped off, then improve that section of the demo. This was studied by Sesame Street creators and well documented in Malcolm Galdwell’s book, The Tipping Point.

One-page sites - This is close to, but not synonymous with Flash sites. This included ajax sites, and clever javascript sites as well as a good portal with the latest trend of the panel boxes (See yahoo.com as an example)

Checkouts / Registrations / Searches - But it’s the opposite here, you want less time, not more. Time here is a factor of how quickly you can process a request. Think of user errors, indecision, pages in the process, render and download times, authentication of credit cards, server loads, etcetera. If the world was buying your book on your eCommerce site, how many could you process in an hour?

It doesn’t count

Established websites - if you are already receiving traffic and you want to make a change and detect the effectiveness of the new design to the old you need to pick large metrics, like conversion actions not time on pages. When you make a large enough change it become impossible to compare apples to onions.

Advanced web surfers - with Cable internet access and not dial-up, with tabbed browsing now on IE7 and obviously on Firefox, along with “restore session” you can hang on to a site not via bookmarking it, but just tabbing to it, and tabbing off to anything else and tabbing back. I’m notorious for a tab bar full of sites I’ve visited over the last 48 hours. This would swing the metric with just one visitor like me into a different world sending the complete wrong signal to the designer and client alike.

Blogs / News articles - traffic is measured by uniques and total visits, or more basically, page requests, or page renders. If I’m buying ads, I want to pay for the page request, not for the 3rd 30 second time slot on a popular news article. Blogs are basically just creative content. The total time alloted to digesting an article depends on a great number of factors, including how bright your audience is, how tiny your text is, how distracted the reader is, if you content is work safe, and on and on. Time means nothing compared to total page views. How big is your audience, not how slow.

Pages with the print button - If you can print it, you can read it offline. Again, blogs, news, even business sites that have details about their product. I don’t mind reading online, but there are some people who can’t stand staring at text illuminated by a big dim flashlight pointed directly at their eyeballs.

I left thinking I didn’t list everything - that I might have to come back and add a few more. But basically, I’m also left trying to defend TOP as a stat, but not as the standing metric. In any design process going forward I will devote some attention up front to whatever stats I’m looking increase and concentrate on getting buyoff and commitment to those metrics to build better designs.