Originally published: May 01, 2001
Web designers often tell us that they spend a great deal of their limited time and resources working to improve their on-site search engines because, they believe, there are some people who always rely on the search engine to reach their target content. They find further support for this assumption from Jakob Nielsen who, in his book, "Designing Web Usability," asserts that more than half of all users demonstrate "search-dominant" tendencies by going right to the search engine when they first visit a web site looking for content.
Assuming this is true, designers have their work cut out for them. Devising and producing a site that supports both visitors who prefer using the search engine and those who gravitate toward links presents a substantial challenge. Teams with limited resources find themselves in the position of having to support two separate paths to the same content. With perhaps thousands of pages of content to get users to, maintaining separate location tools becomes a monumental task. Anything we can do to reduce the work is going to be tremendously appreciated.
So, we decided to put the user search-dominance theory to the test. In a recent study we conducted on e-commerce sites, we tested out this assumption about user preferences.
In our study, we observed 30 users performing 121 different shopping tasks. Each user visited between 3 and 6 web sites, shopping for items they told us they were interested in purchasing; no two users were interested in exactly the same products.
If the search-dominance theory is true, we should've seen a subset of our users always relying on the search engine to find product information, while others relied on the links. If we didn't see at least a few users who consistently relied on the search engine, then we would have to question this idea of search dominance.
Also, when we looked individually at each site in our study, we should not have seen all the users who visit a particular site employ a single strategy. We expected to find samples of each kind of user behavior on each site.
To make this last point a different way, consider the city or town where you live. Some of its inhabitants are right-handed, some left. If you go into any restaurant in town on a Saturday night, you should find some mix of lefties and righties. It is highly improbable that only right-handers would populate a random restaurant on a random Saturday. We had a similar hypothesis for these web sites: It seems highly unlikely that only search-dominant users would use a site during a given series of tests.
When we looked at the data from our study, we found that there wasn't a single user out of 30 who always used the search engine first when looking for product information. None of the users in our study were search dominant. However, we did uncover some link-dominant users. About 20% of our participants chose links exclusively.
Then, when we looked at the individual sites, we saw that for 21% of the sites, every single user who visited only used search. It seems that these sites were search dominant, not the users. Thirty-two percent of the sites were link dominant (users only used the links on the site) and 47% were not dominant to search or links.
We find it fascinating that on 53% of the sites we tested, each visitor stuck with a single location strategy the same strategy employed by all the other visitors to that site. This implies that there is something inherent in the site's design that causes users to choose the search engine or the links, not a hard-and-fast preference of the user.
As we talk with the users, we often hear them tell us that they do have a preference for search that they are search dominant. All the time, we hear, "I always go to search immediately." But none of our users actually did always go to search immediately yet another piece of evidence to suggest that what users say they do and what they actually do are very different.
In analyzing our data, we noticed that one of the factors that predicted whether users would initially start with search or with the links was the type of product being sold on the site. Certain types of products lend themselves better to being searched. For example, users typically go to the search engine to find a specific book or CD, however they tend to use the links to find a particular item of clothing. We believe that the nature of the content on the site can play a huge role in whether it is a search- or link- dominant site.
We also noticed that users often gravitated to the search engine when the links on the page didn't satisfy them in some way. For a long time, we've observed that users seem to use the search engine as a fallback after failing to pick up "scent" on the home page. Our recent study gave us more evidence to support this behavior. We observed many home-page link failures that forced users into the search engine.
The lack of evidence to support the user search-dominance theory implies that teams may need to think about concentrating their efforts on a single location tool. Depending on the specific content on their site, teams might want to focus specifically on either the search engine or the links, but not necessarily both. Everything we've seen in our testing says that focusing the resources on a single approach can dramatically improve the user's experience.
Read related articles: