Jakob Nielsen's Alertbox, June 23, 2008:

Extreme Usability: How to Make an Already-Great Design Even Better

Summary:
The 1% of websites that don't suck can be made even better by strengthening exceptional user performance, eliminating miscues, and targeting company-wide use and unmet needs.

99% of readers can stop here; this column is for people who have a great website or intranet.

More typically, your site sucks and it's incredibly easy to discover your main usability problems. Test with 5 users and you'll likely get enough insight to double your site's business value. This step is easy; we can teach design teams to run simple usability studies by walking them through a complete test of their designs in just 3 days.

But what if you've already run many rounds of qualitative user testing and improved your site through iterative design? Say you doubled your conversion rate. And then you doubled it again. You're now at a stage where you have a really nice design that satisfies all the obvious user needs and lacks obvious impediments to use. (Like I said, I'm writing for only 1% of readers here. But eventually, you'll be in this situation if you keep polishing your website or intranet through continuous quality improvement.)

Once you've picked all the proverbial low-hanging fruit, it's no longer so easy to improve usability. Quick simple studies are not enough; your design's remaining issues won't be of the gobsmacking kind that are glaringly obvious from testing a handful of users.

To make a good user experience even better, try the following four ideas:

• Identify where your design exceeds expectations, and apply that success even more broadly.
• Look at things that are close to going wrong and ensure that they never will.
• Go beyond user experience for individual customers to consider enterprise usability.
• Discover unmet needs.

When Things Go Well

Taking a metaphor from the airline industry, it's now time for you to study "safe flights." Historically, the airline industry improved safety by studying airline accidents. Modern airplanes now crash so rarely that the industry studies safe flights to achieve further improvements, investigating why flights are safe and whether they had any close calls.

Similarly, you can now review usability study segments that you often skip — that is, those sections where users immediately click the right thing, have no difficulties, and basically like your site. Why was the correct link the most apparent? Why was it so easy for those users to accomplish their goal? And what did they particularly like about the design?

You likely have site areas that have yet to implement all such best practices. Hunt them down and elevate them to the standard established by pages that users love.

To exceed the average levels of user success, identify "lucky" users — the ones who are outliers in terms of uncommonly successful, high-performance use of your design. Do they exhibit any special behaviors that you might strengthen and make even more fruitful? Are there ways to encourage average users to exhibit those same "lucky" behaviors so that they can benefit as well?

Tiny Symptoms Pinpoint Incremental Opportunities

We often observe users who hesitate before clicking or initially misinterpret information, but then correct themselves before they get into trouble. In such cases, it's tempting for the design team to simply utter a collective "whew" — we almost had a usability problem, but our design won in the end.

Certainly, if your design ultimately works for your test users, you don't have high-priority usability issues. But miscues are still symptoms of something you ought to correct (once you've fixed the design disasters, of course). Correcting miscues is important for two reasons:

• A miscue wastes time. A hesitation might last only a second, and a misinterpretation might be cleared up in five seconds. But that's still lost time. For a website, slower user performance translates into less-than-smooth user experience; the site just doesn't quite feel right and thus reduces customer satisfaction. For an intranet, every wasted second lowers employee productivity. And, because miscues happen hundreds of times per day, they can add up to a full workday over the course of a year.
• Some users are less lucky. Even if all your test users overcome a miscue without getting into real trouble, some users will in fact click the wrong link or act on their misinterpretation. We can't test millions of users; anything that's error-prone in a test will, in real life, trigger an actual error in a small percentage of cases. If your design is only slightly error-prone in user testing, you'll observe miscues as symptoms rather than seeing actual mistakes. Real errors will still happen, just not very frequently.

If you conduct eyetracking studies of your content, you'll sometimes identify regressions: cases in which the eye moves back instead of forward because the user needs to read something twice.

A regression is another symptom of a small problem: the text was difficult to understand. Regressions can result from several different readability problems:

• A word that the user didn't know
• An overly complex sentence structure
• A muddled argument or seemingly dubious information

The vast majority of websites and intranets don't need to worry about regressions because they've yet to rewrite their content to comply with the existing guidelines for writing for the Web. In such cases, doing eyetracking studies is a waste of money. But once you've achieved good online writing, watching for regressions can help you polish content usability even further.

Enterprise Usability

Most usability methods focus on improving the user experience for individual users. However, there is a bigger scope of experience: enterprise usability, which considers a design's implications across an entire company.

Improving experience beyond the individual level is particularly important for B2B sites. A classic business example is the need for content that can help users convince an ultimate decision-maker about the value of a product or service. In B2B, it's often not enough for users to think your product is good — the person charged with researching purchases on the Web might need to make a presentation or write a report to a committee or a boss who will make the final decision.

To study these organizational issues, you need broader research methods, which are more expensive than single-user testing.

Discovering New Needs

No matter how good you are at meeting customers' current needs, there are probably opportunities to do new things that they don't even realize they want. If you discover something truly new, you can even patent it and thus ensure a sustainable advantage over your competitors.

How do you identify needs that customers don't know they have? By running field studies in which you observe user behaviors. As I've said a million times, don't listen to what they say; look at what they do.

Sadly, field research is much more expensive than testing a handful of users in the lab. But, once you've perfected your current UI, that's what you have to do to advance to the next level.

Are Small Improvements Worth the Effort?

Most websites and intranets offer such a miserable user experience that it takes only a microscopic usability effort to create astounding improvements. The return on investment is usually several thousand percent for a company's first usability project. ROI continues to be sky-high for many years, because it takes a long time for a company to progress in usability maturity to the stage where it routinely releases designs with superb user experience.

But once you're good, then what?

ROI will be smaller when more usability work is required to identify fewer design improvements. Still, I have two arguments for why you should nonetheless continue usability work:

• Small ROI doesn't mean negative ROI. Obviously, if you lose money on usability, you shouldn't do it. But I don't think any website is so perfect that further improvements would be worthless. As your site improves, usability might become less of an urgent priority, but it shouldn't be abandoned.
• Competitive pressure means that this year's leading design will be an average design in a few years — and a miserable design a few years after that. User expectations continue to increase. Thus, even if you could afford to halt your usability efforts today, you'd have to resume them in a year or two as the competition passed you by. But, since it takes several years to build high-level usability competency in an organization, you can't simply resume usability at will after it's been discontinued. If you really think it will be 2 years until you find any further design improvements worth doing, spend the interim doing fundamental user research into your customers' deep needs.