Interview with Jakob Nielsen

Matt Mickiewicz

Matt Mickiewicz interviews Jakob Nielsen, author of the brand new book "Prioritizing Web Usability," about AJAX, usability’s close link to keyword advertising, and some of the advertising formats we’re seeing around the Web today.

What do you think of the current implementations of AJAX
(Google Maps, Writely, Google Suggest, Zimbra, etc.) on the Web today?

They are irrelevant for the vast majority of business web sites. (And by "business web site" I also mean sites for government agencies and non-profits.)

A business site will profit much more from writing better headlines than from sticking a programming trick on its pages. Take maps: yes, it’s one of the usability guidelines for store finders and locators to show a map of your location. But don’t make this a highly interactive map — it’s better to use a hand-drawn map that’s optimized to giving directions to your address. One that shows landmarks, recommended parking lots, and bus stops, and emphasizes freeways, but excludes the small side streets that people probably wouldn’t use when driving to your building.

Similarly, when we tested a large number of investor relations sites, we found that advanced tools for plotting stock trends and financial numbers only confused most individual investors. A better alternative is to show the most important information in a static plot that’s been optimized by a good designer. (Yes, institutional investors in our test did use advanced visualization tools, but they did so on their Bloomberg terminals. On a company’s own IR site, they were looking for the CEO’s vision for the company’s direction.)

Admittedly, Google Suggest is a neat hack. I think similar ideas would be useful on extranets when users are filling out form fields restricted to a few thousand possible entries, such as product names or destination addresses for a shipment. Type-ahead could save a lot of keystrokes on sites that are used repeatedly.

The definition of usability states that usability is a combination of five quality attributes, including learnability and efficiency of use. You have to balance these qualities differently depending on the type of user interface you’re designing.

It’s important to remember that most web sites are not used repeatedly. Usually, users will visit a given page only once. This means that the efficiency of any given operation takes a back seat to the discoverability and learnability of the feature. Therefore, interaction techniques like drag-and-drop should almost never be used on web sites. Instead, focus on showing a few important features and making them accessible through a single click on a simple link or button.

Some business sites that are used repeatedly include features for approximating software applications. Online banking comes to mind, and I can easily envision a design that enables the user to see the front or back of a check through an AJAX technique on the account statement page, instead of going to a new page.

The very nerdiness of the name "AJAX" gives me hope that it will be used for causes more worthwhile than those now characteristic of Flash. Doomed by its own name, Flash had similar potential but was so grossly abused for "flashy" design that it never succeeded in adding much useful functionality to the Web.

Since AJAX is an engineering idea, it has the opposite problem of leading to featuritis. Remember: just because you love technology and advanced features, it doesn’t mean that your customers do. They just want to get in and out without worrying about your web site. So take it easy on the features.

In your book "Prioritizing Web Usability" you say that future increases in keyword advertising will be tightly related to usability. How so?

This is because better usability means higher conversion rates for a web site. The more visitors you convert, the more you can afford to pay to attract a visitor in the first place. This again means that you’ll bid up the keyword auctions as you improve the usability of your site. Assuming that at least ten companies in a certain business sector all improve their sites, the end result will be higher keyword prices.

So improving usability and improving conversion rates are closely related?

In principle, the two are different. Usability is one of the key influencers of the conversion rate, and improving usability will typically double the conversion rate for a site that has had low usability in the past. But there are many others factors that influence the conversion rate, including the brand’s reputation, the product itself, and obviously, the price. Lower the price and you will almost always increase the conversion rate — but of course you will probably lose money, so it’s better to invest in the other factors.

I’ll give you an example from my own company: we sell a report with usability guidelines for email newsletters. The previous edition was 293 pages. The new edition is 544 pages because we’ve added findings of how people read email from a recent eyetracking study. Even though the price increased from $298 to $398, the new edition is selling much better than the previous one.

I don’t think the increased sales are caused by a lower price per page. I think they’re a result of delivering content that matches the customers’ needs better. Companies that publish email newsletters don’t have a lot of places to turn to for usability advice, and are therefore more interested in highly detailed and thorough research.

In this example, we’re getting a higher conversion rate for the revised product, despite a higher price and no changes to the web site. It’s the product itself that’s the cause of the increased sales.

Of course, it’s not particularly revolutionary advice to tell companies to make products that customers want, but it’s worthwhile advice because it’s the number one way to increase sales. And it’s important to remember that user research is one of the best ways to find out what your customers want before you go to the expense of building the product. User research shouldn’t just be used for web design; it should also be used to determine your product strategy.

Given that price, product, brand, and usability all impact conversion rates, it becomes a question of where to invest your resources. I certainly wouldn’t claim that usability should get all the money, but most companies under-invest in this factor.

Consider how expensive it is to run a big advertising campaign: you can easily spend millions of dollars. In contrast, a simple usability test costs $25K — less if you do it yourself.

Also consider whether it’s worth wasting money to drive traffic to a web site that’s so bad, users leave after a single glance. Unless you’ve done your homework and ensured decent usability, you shouldn’t invest in any promotions.

A lot of search marketers are investing heavily in A/B split testing to maximize conversions from paid search engine traffic. When it comes to increasing conversions and maximizing usability, what do you think the proper role of A/B split testing is? Can it ever replace a live test?

A/B testing can never replace live user testing for the simple reason that it doesn’t give you the crucial insight into why users do what they do. You need the "why" in order to generate design insights that can set the directions for the next steps of your project.

A/B testing is very narrowly targeted at deciding which one of two designs generates the highest number of desirable actions, as long as these actions can be measured. This means that:

  • you need to have designed and implemented both alternatives
  • the outcome can be counted by a computer

In many cases, a web site’s worst problems are not issues that you’d account for in an A/B test because you wouldn’t be aware of them unless you’ve carried out usability testing. When you limit your research to the ideas you can generate yourself, you have closed your mind to all the unexpected things that users do. In usability tests, the issues you didn’t consider during the planning process usually turn out to generate the biggest findings. That’s why it’s so important to set users loose with a task, and sit back and observe their actions.

When a company finally is ready to invest in promotions, what’s your opinion of the usability of the syndicated ad formats, like Google AdSense, that are so popular on the Web today?

Our eyetracking studies show that users do look a little bit at these text-only ads. Probably because the syndicated ads look very similar to the search engine ads that people are used to seeing. Users have discovered that search engine ads tend to present fairly useful information, so users frequently look at the ads on a SERP.

Sadly, the syndicated ads are not nearly as useful to users as search engine ads, which is why they tend to have much lower clickthrough rates. There’s a big difference in user intent between doing a search and browsing a content site. On the search engine, the user is primed to go someplace else, and the ads can show offers that are pretty much exactly what the user is looking for. In contrast, an ad on a content site might look the same as a search ad, but it doesn’t offer the same level of specialization, and thus it doesn’t have the same effect.

An even newer trend in advertising is the contextual text link. In this format, certain keywords are associated with, and hotlinked to, specific sponsors. What’s the average user’s perception of those ads, and should advertisers be buying into this new format?

"Disgusting" is the only appropriate term for these ads.

They are undermining the most central element of the Web’s user experience, which is the understanding of what a blue underlined word stands for. Hint: it doesn’t stand for an ad. Luckily these ads are found on very few web sites, so they haven’t confused users that much yet. But if the ads were to become prominent, users would become much more reluctant to click on links, which again would drastically reduce the usability of the entire Web.

When visitors do finally arrive on a web page, you’ve noted a huge disjoint in the number of users who will scroll on a content page (42%) or a SERP (47%) vs. the homepage (23%). Why do you think this is?

In the studies we describe in the book, homepages were scrolled much less than other pages for two reasons.

First, many web sites actually do a pretty good job at putting the most important features or biggest news items at the top of the page. Thus, users often don’t need to scroll.

Second, users don’t want to engage with a homepage. They’re on a web site for a reason, and that reason doesn’t include finding out about all the site’s offerings. People are very fast at moving off the homepage because they want to get to the part of the site that addresses their current concern. Content pages and SERPs get scrolled more because they tend to be focused on a problem that’s specific to the user rather than providing a generic overview of twenty different things.

In your scale of usability misery on the Web today, site searching is now the worst offender. What’s the easiest solution for an overworked and overburdened Web Developer who wants to improve his or her site’s search usability?

The easiest solution is to subscribe to an outsourced search, and delegate the problem to somebody else. I don’t necessarily think that this is the best solution, but it’s definitely easier than installing high-end search software, and then tweaking it to suit your circumstances. If you don’t customize the search software, you’ll end up with a miserable search, in which case it’s better to use an optimized solution from a company that specializes in site search.