Originally published: Jul 01, 1998
Users say they dont like to scroll. As a result, many designers try to keep their web pages short.
But one of the most significant findings of our research on web-site usability is that users are perfectly willing to scroll. However, theyll only do it if the page gives them strong clues that scrolling will help them find what theyre looking for.
Designers have two choices when deciding how much to put on a web page:
Our research shows that this second strategy fewer, longer pages may be the best approach for users.
In the trade-off between hiding content below the fold or spreading it across several pages, users have greater success when the content is on a single page.
Many users apparently choose links by first eliminating the links they dont think will work. If designers spread information across several pages, they force users to "pogo-stick" back and forth between links and other pages a far less successful strategy. (See "Pogo-sticking: Downs and Ups" in the July/August 1997 Eye For Design.)
Users may tell us they hate scrolling, but their actions show something else. Most users readily scrolled through pages, usually without comment.
On the version of the Fidelity site we studied (fidelity.com), the designers had clearly decided to minimize scrolling by building short pages: no page extended beyond a single browser window. We cant be sure users even noticed this feature: when we asked them what they liked about the sites, not one commented about not having to scroll on Fidelitys site. Compared to sites where users had to scroll, this site did not do significantly better or worse.
We asked users to visit a web site and answer specific questions, assuring them that the answers were on the site. We noted every link they clicked, then later measured its distance from the top of the page. Users who chose links from farther down the page were more likely to succeed.
Users who chose links closer to the bottom may simply have been more thorough in their decision-making. But it turns out that links deep in the page have other interesting attributes: they are mostly text links instead of graphical links, they tend to be longer, and most of them are content links pointing directly to content.
Near the top of the page, category links which point to other links are more common; weve found that content links are the most successful type. (See "Not All Links are Created Equal" in the September/October 1997 Eye For Design.)
One criticism of long web pages is that they hide some information, forcing users to scroll. Short pages may avoid this potential problem by showing more (or all) of an individual page, but the information is still hidden on other pages. Users must still click repeatedly to get to the desired information.
Designers can easily control one theoretical disadvantage of longer pages the difficulty of choosing a link by introducing a structure that users understand. Users make better link choices when they can more readily eliminate "wrong" links; organized links help them do this.
Although scrolling requires some effort from users, so does looking at a page and deciding which link to click. Grouping links seems to help.
Users were more successful in finding information on pages with many groups of links at least up to 10 groups. (We have less data on pages containing 10 to 20 groups, but our analysis found nothing to suggest that larger numbers of groups are harmful.)
One structure that seems particularly useful is placing similar links with each other; this seems to help users discriminate. Finding nearby links they recognize may have assured our users they were in the right place or close to it. Groups of links may also make it easier for users to exclude links that seem less likely to succeed. With less effort required to zero in on a link, users may be more willing to expend their limited effort budget by scrolling.
Grouping links is only one way to structure information. In our analysis of successful web pages, we also looked at factors such as alphabetized lists and levels of information. We found that users quickly seemed to grasp that a list was in alphabetical order and would scroll to find the item they wanted.
Increasing the levels of information similar to adding sections to an outline also seemed to help users. Both these strategies produce longer pages with more links.
The users we tested were often willing to scroll long web pages, but a couple of design factors may have stopped them prematurely.
One page from the WebSaver site (websaver.com) has four groups of links in no discernible order. Nothing suggests that users will find more information further down the page. Our users sometimes stopped scrolling when they reached one of the horizontal lines demarcating areas of the page.
Horizontal lines like these may subconsciously cue users that theyve reached a stopping point. The presence of horizontal lines wasnt statistically significant in our analysis (possibly because we didnt test many pages with horizontal lines), so this is only a hypothesis. However, we saw this stopping behavior several times.
On the other hand, users readily scrolled a page from the Edmunds site (edmunds.com), despite the many horizontal lines. We believe users quickly grasped the fact that the manufacturers names are listed alphabetically, implying that theres more information farther down the page.
A row of text links may also be a subtle "end of page" cue.
In the version of Boston.com we tested, the first screen of the home page included several graphical links followed by some text links. Text immediately below these links told users to scroll for more information, but none of our users did. This could be because pages often end with a navigation bar of text links at the bottom. Because copyrights and credits also frequently end a page, their appearance elsewhere may have a similar effect.
Read related articles: