Jakob Nielsen's Alertbox, February 6, 2006:

Users Interleave Sites and Genres

Summary:
When working on business problems, users flitter among sites, alternating visits to different service genres. No single website defines the user experience on its own.

To get the big picture, you often must delve into the details. In several recent studies, I've observed an important shift in user behavior, but simply stating the conclusions upfront would obscure the true business impact. To really see what these changes mean for individual sites, we must look at users interacting with specific businesses.

As a case study, let's examine the detailed behavior of a user from our current B2B usability study. The user was researching the purchase of a portable computer projector for her boss to use during presentations. In the B2B sector, this is a somewhat low-end product, running about $1,500. However, this example's low cost and relative simplicity make the user session representative for much user behavior in the higher end of the B2C sector. (In most B2B study sessions, we observed highly specialized personnel researching advanced products or services costing tens or hundreds of thousands of dollars. They're thus less instructive for non-B2B readers.)

The sidebar shows a step-by-step transcript of the user's progress through the different sites she visited while researching the projector. This detailed information is worth reading; it gives a vivid picture of how much overhead the user had to suffer to get a small amount of actionable information.

Time Spent and Sites Visited

In the total elapsed time of 44 minutes and 25 seconds, the user had 93 pageviews across 15 different sites. On average, she spent 29 seconds per pageview. This is close to the mean in our study of user behavior on 25 mainstream sites.

At the session's end, the user had a shortlist of three projectors to take to her boss, with a recommendation to buy the Mitsubishi XD480U. Although it was more expensive than the other two, she thought the Mitsubishi's longer lamp life was worth it, given the high cost of replacement lamps.

In total, the user issued thirteen searches -- ten using Web-wide search engines and three using specific websites' site-search features. Of the fifteen sites she visited, the user got to six by guessing their URLs (two of these she guessed wrong, but later found through searching; the other four she accessed directly by typing in the URL). She visited one site through a link from another site; she visited the remaining eight by following search listings (six organic links and two sponsored links).

A heavy use of search is typical of what we see in many studies. In this case, the test user was very knowledgeable, which is common in B2B. She had bought many projectors before, though they were stationary models. Because of her previous experience, she was able to directly type in more website URLs (and thus go directly to sites) than is typical in most of our studies.

The user's sponsored link clicks were disappointing from an advertiser's standpoint: she wasn't looking to buy on their sites, merely to look at the information they provided. For example, after having identified several possible projectors, the user wanted to estimate total cost of ownership by adding in the cost of replacement lamps. None of the vendor sites provided this information, so she searched for lamp prices and clicked on an ad from an e-commerce site that specialized in lamps. The site easily helped her complete the TCO estimates, but she wouldn't likely make a lamp purchase until after the projector had been in use for several years. It's highly doubtful that she'll remember the lamp site years later, so it didn't get any business from paying for the click.

Cutting Across Genres

The following pie chart shows the distribution of time the user spent across the fifteen visited sites:

Each slice indicates the time spent on one site;
the slices are color-coded by the type of site.

Even a single session provided plenty of specific usability findings on the individual sites that the user visited. We're combining these insights with our observations of other users to derive comprehensive design guidelines for B2B vendors, B2B e-commerce sites, and B2B content sites.

Here, I want to focus on two broader points. These points are critical for both B2B and B2C, and they've been validated by many other studies:

Users flitter between sites. Users gather snippets of information from multiple, competing sites to build a thorough understanding of a specific problem.
Users interleave visits to different site genres. Users alternate between vendors, retailers, and reviews, using search engines as a hub.

In our example case, the user also integrated offline information sources. For example, none of the sites helped her figure out how many lumens were needed to show presentations to her target audience size. Luckily, our user was an experienced projector-buyer and owned a reference book with this information. She also relied heavily on written notes to record and remember information from one site to the next.

Assuming that you can design the entire user experience is a lost cause when your target users are engaged in non-trivial tasks. Almost all B2B -- and many B2C -- user experiences involve multiple sites. Two conclusions for site owners:

Jakob's Law -- "users spend most of their time on other sites" -- now has a corollary: Even during a single task, users often spend most of their time on other sites. In our example, the three sites that the user spent the biggest percentage of time on (CNet, Google, and Dell) each accounted for no more than 17% of the session. Again, this emphasizes the importance of complying with usability guidelines and design conventions. Doing so lets users focus their brief site visits on discovering your products and services, rather than struggling with your navigation or trying to learn non-standard features.
Provide all the information users need. Fail this and you're inviting users to leave, which they'll do soon enough anyway. In particular, your site should show prices (violating this guideline was the #1 Web-design mistake of 2002). If users need to search out other websites to get their questions answered, they'll likely discover alternative solutions to their problems elsewhere rather than explore your site's options. The more clumsy and less informative a site seems, the faster users will abandon it.

When you watch people use the Web as a whole (as opposed to testing an individual site), it's striking how much crud they have to wade through. Our user openly mocked sites that slowed her down with internal ads or banners proclaiming their products as "new & exciting." To improve the business user's experience, follow three simple design guidelines:

Focus on products. Include them on the homepage and get people to specific products quickly. The more time it takes users to find the first answer, the less likely they are to look for further answers on your site. Many other sites beckon.
Use dimensional navigation to link from individual products to others that are similar, but differ on an important target dimension. For example, link from one projector to another that has the same luminosity but lower weight. This encourages users to explore your product line.
Link to other sites that show you in a positive light. People will leave anyway, so you might as well steer them to a place that does you good. Otherwise, you're at the mercy of whatever their next search dredges up.

Web-Wide Improvements Needed

The Web failed our user in two ways:

There's no support for multi-site behaviors, beyond the minimal integration that search engines offer and a few cases of review sites and vendors linking directly to e-commerce sites' product pages. Our user typically had to build her understanding of a given projector by combining information from four sites: the vendor's site, a review site, an e-commerce site that sold the projector, and an e-commerce site that sold the lamps. To improve the total user experience of the Web, we need tools that cut across individual sites.
The user never found good independent reviews of some products she was considering. There are two primary reasons for this. First, it's a failure of the Web's current business model: there's not much incentive to produce independent content when search engines earn far more revenue than the actual content providers. Still, vendors are losing significant business opportunities by not linking to informative reviews on independent sites. (Doing this is a long-standing e-commerce usability guideline, but it surely applies to B2B vendors as well, even if they don't sell from their sites.) Second, search tools fail to pull up relevant reviews because they fail to match specific classes of user needs with their associated content. The current model of "understanding" content purely in terms of keywords and generic quality ratings is insufficient to support user needs.

It's a given that most users view the Web as a unified resource and combine sites for many of their tasks. Neither browsers nor sites have caught up with this new user behavior. Both will need to change in the coming years.

Learn More

282-page report on B2B usability with 144 usability guidelines and 158 screenshots is available for download.

Full-day seminar on B2B Website Usability at the User Experience 2008 conference in Chicago and Amsterdam.

We'll discuss ways of analyzing test findings and user behavior in further depth in the full-day tutorial on user testing.

> Other Alertbox columns (complete list)
> Sign up for newsletter that will notify you of new Alertboxes