In the course of dozens of discussions at conferences and in social networks, I continue to find myself debunking two popular myths:
- If you force people to search for information, you have done a poor job with your information architecture (IA).
- Nobody reads on the web.
I say they are myths because I have reason to believe that they are not generally true. As far as I can tell, they result from user experience testing on small groups of users in controlled environments. When I examine these tests, I find two of the common fallacies of social science:
- Drawing sweeping conclusions from small samples of homogenous users
- Failing to control all the variables
Despite my skepticism, I held onto these myths until I started seeing broad-based evidence to the contrary:
- In surveys all over the web, users show an overwhelming preference for search as the primary way they find information.
- On multiple sites within ibm.com, users showed a preference for reading over other information tasks, as long as it was good content.
I have spent the last five years of my career trying to debunk the myths for which this evidence stands in stark contrast. Why? Because I am in the business of helping IBMers create more effective websites, and these myths are getting in the way of their success.
Myth # 1 leads to experiences in which you surface complex taxonomies that users have no idea how to use. If you try to minimize search, you maximize navigation. But most sites just have too many options for users to wade through and navigate comfortably. By shirking search, you overwhelm users with navigation options.
Myth # 2 leads to design-heavy experiences devoid of content that neither users nor search engines can effectively parse. Search engines use text to decode the relevance of pages to queries. If you think users don’t read, you don’t provide search engines with enough text to determine relevance. It turns out users are confounded by the same problem as search engines. This should not be a surprise because search engines are in the business of giving users what they want, and search engines favor pages with a healthy amount of text.
In the last five years, I have built a compelling set of arguments to debunk these myths. I’d like to share them with you now. The best way to debunk these myths is to violate them and see what happens to your metrics. With the help of our book, Audience, Relevance and Search: Targeting Web Audiences with Relevant Content, I have managed to convince people at IBM to build experiences that go against the very premises of these myths. We have gathered a growing pile of evidence that not only are the myths false for some users, they are false for most users.
Users Prefer Search to Navigation
A few years ago, we began a project at IBM to upgrade our ibm.com search engine. Our user satisfaction surveys consistently pointed to search as the prime culprit in user dissatisfaction. And metrics backed this up: We had horrible abandonment rates that, over time, led to a lack of usage of what should have been our site’s most popular feature.
I say it should have been our most popular feature because our home page contained more than 200 persistent links to product categories, solution areas, and major sections of our site, such as documentation and support. Psychologists of language have a rule for how much an individual can understand at a time: seven plus or minus two. No human can comprehend more than nine discrete linguistic items at a timenine bullets, nine major headings, etc. Many can only comprehend five. And we were presenting 200 discrete linguistic items to our unsuspecting users!
When these users fled to the relative safety of search, it only got worse. They were presented with thousands of items, many of which appeared to be duplicates. Oddly enough, the most popular feature on our search engine page was an ad-like function that brands could put on the site to usurp search engine results. Because the brands would only pay for relevant listings, the ads were more relevant than the organic results. Our manual efforts were more effective than our automated search engine.
I helped write the business case for the new search engine and was on the development team that built it. Within days of launch, our abandonment rates went way down, which we expected. Also, organic usage all but eliminated paid usage, which we also expected. Then something happened that we didn’t expect: Search usage started to go up. Month after month, as users learned to trust our search engine, they started using it again. Prior to launch, it was the 20th largest referrer to our pages. Google was the first referrer, with Yahoo and Bing in the top five. A year after launch, it was our second largest referrer, and gaining on Google.
The myth that users hate search sprang up when search experiences were horrible. If you provide good search experiences, users will choose them over navigation 9 times out of 10. Good search experiences are simply the fastest way to find what you’re looking for when you know what you are looking for. How many users of corporate websites don’t know what they’re looking for?
My data only corroborates a growing ream of evidence favoring search. The preference for search is growing as more users access the web from mobile devices, which simply don’t have enough screen real estate to provide many navigation options. I think we can safely bury the myth that users hate search. It’s time we embrace search and begin building experiences expressly for search.
Users Like to Read Good Content
I’ll admit I held onto this one longer than I did to the search myth. As far as I can tell, it stems from a series of usability studies by Jakob Nielsen and others, which suggest that users primarily scan and skim on the web. I am a great admirer of Nielsen, but I had my doubts. After all, content is king, right? The web is primarily a text-based medium. How could it be that users don’t like to read?
It turns out the myth did not match our metrics, at least not consistently. We had lots of pages that people appeared to scan and not read. But we had many more experiences where people seemed to read and consume readily. developerWorks is a good example of the latter: very popular content and very text-heavy.
It didn’t take me long to be able to put these two groups of pages into discrete buckets: those that users scanned and abandoned had very little content. And what content they had did not provide any real value. It was just generic, anonymous hype about product features and benefits without any proof points or concrete case studies. The pages people seemed to like didn’t say anything without evidence and made compelling reading, in part because they focused on real people solving real problems. Could it be that it was not the web that made people averse to reading but the content on the web? Duh!
Shortly after this “duh” moment, I read an unpublished report from our Cognos brand. Cognos had a long history of web content excellence prior to being acquired by IBM. After acquisition, the new IBMers wanted to try to instill some of the same best practices into IBM content. They showed me slide after slide of cases where the most popular links on the page were the Next or buttons at the end of long pieces of copy.
They started out building pages with very little copy, following Nielsen’s advice. But they let their metrics lead them in the directions their users wanted to go. After five years of regular page updates, the clear choice for most of their users was to read. Apparently, the Cognos web development team had very good writers and editors.
Since that report, I have tested this concept everywhere I have been involved in content development. When I worked on the Smarter Planet experience, we built rich articles for users on most of our pages, such as our Water Management topic. These were some of the most popular experiences we gave them.
On the IBM Cloud Computing site that I now work on, we built a long piece of content with a Read More button in the middle of it. That is one of the most popular calls to action on the site. Everywhere I turn, good content gets rewarded with strong engagement from web users looking for something to read.
The myth that users don’t like to read on the web is not generally true. Though web users are time- and attention-starved, I suggest that it’s not because of the web. It’s because the content presented to those users in studies was not compelling enough for them to want to read it. Or perhaps the search results in 2005, when Nielsen first published his studies, were typically so poor that users scanned and bounced more often than they scanned and read. For whatever reason, in 2011, if you provide relevant, compelling content, show don’t tell, advise don’t sellusers read it.
The two myths debunked in this article are related, of course. If users prefer to search and they like to read (as long as the content is relevant and compelling), the first order of business in web publishing is building relevant content to the search queries your target audience uses.
Saying users like to read does not imply that they don’t scan before they read. But scanning is the way users determine page relevance. If you give users strong visual cues that your page is relevant to their search queries, they will begin to read your content. If it makes compelling reading, they will engage deeper into your site, taking your other calls to action.