Web Site Expert巻頭レポート(英語)

Jared Spool on the Craft of Good Interface Design

Jared Spool has been doing usability studies before the term⁠usability⁠⁠ was invented. He founded his Massachusetts firm, User Interface Engineering, back in 1988. Early projects included helping design teams evaluate prospective user interfaces for airplane cockpits, medical equipment,and molecular modeling tools?all technologies where poor interface design can have catastrophic consequences. Indeed, Spool says that the modern age of interface design came out of the 1979 Three Mile Island nuclear power plant accident, in which a flawed control panel led to operator errors.

On the software side, Spool saw a paradox in the friendlier interfaces made possible by the continued doubling of processing power predicted by Moore's Law. But as interfaces grew smarter, the people using those interfaces were less technically savvy. Today, websites represent a microcosm of that problem?a place where people, with a few clicks of the mouse,can book a vacation or pay a bill, but also reserve a hotel room on the wrong side of the country or overpay their electric bill ten-fold. Spool argues that on the Web, bad design is more the rule than the exception.

User Interface Engineering's own website resembles not so much a design firm as a think tank, with an active blog, an archive of podcasts, Twitter feeds, and research reports. Behind it all is the assumption that website design can only be done successfully by design teams who are keeping up with the field as diligently as doctors should keep up with medical advances. The corporate design culture also matters: it should not only encourage success, but permit failure.

Spool is a strong believer in data, but he does not think much of eye-tracking studies, and argues that analytic tools only give you part of the story. He is a big fan of talking to actual users. That means testing your design with the people who actually use your site, listening to what they say,looking at what went right and wrong, refining your design accordingly?then beginning the cycle again. More than inspiration, says Spool, persistent hard work is the secret to a successful design. Web site design may not be nuclear science, but it is a serious business nonetheless.

What is your background?

I started as a software engineer and worked on the first voice mail and email systems. On an American PC keyboard, there are six keys that do insert, delete, page up and down, home and end. I designed the original set.

We started as a consulting business working with software clients like Microsoft and Lotus, and financial firms like Fidelity and Schwab. Our contribution was to help large organizations make sure that the things they build can actually be used by their intended audience. As we were doing that, we realized we were seeing the same problems over and over again. So we began looking into ways to prevent problems from getting into the design in the first place. We began cataloging these patterns, and that became the core of our research operation. Tracking these patterns remains one of the most interesting,lucrative parts of our business.

What kinds of things are you researching?

One of them is the link, which is one of the distinguishing elements of the Web. It's an essential navigational device, and most websites have hundreds,if not thousands, of them. But where have you seen any conversation about what makes a good link good? For example, what do you put in a link if you want someone to click on it, or if you want people to click on it only some of the time, or if you want them to understand where that link will take them? We've been studying this for 12 years now and have developed a solid understanding as to what makes a link good.

More recently, we've been looking at website development teams,themselves. We're interested in how people create things and what the process entails. We look into usability problems after they've occurred and ask if there was away to prevent them in the first place. We've been looking at what teams do who create great experiences. One of the research questions that we've been studying for about eight years now is this: if you want your product or service to be the next iPod or Netflix of your industry, who would you hire? What would you ask them to do? How would you tell if they are doing a good job? What tools do they need? What type of delivery schedule can you expect? What resources do they need? We've made some solid progress answering thesequestions. And we're now working on a book.

From Ajax and beyond, the Web has seen a lot of new technologies since you began. Have they changed the essential design problem?

Actually, the essential design problem remains the same. Let's go back to the link. A link is essentially a stimulus to the human brain to create a muscle movement, which is to click on the mouse. A stimulus is really a neuro-physical response mechanism?and those have remained largely unchanged for the past few thousand years. So at its core, interface design is really about trying to better map to that neuro-physical response. That is, we want to understand why people click, or don't click, on a link?and do a better job of fulfilling their expectations along the way. Everything we are seeing in terms of Ajax, tags, and other ways of representing information online is really about the evolution on the design side, not the evolution on the user's side. The core problem remains the same: find better ways to design interfaces that work the way people work, rather than the other way around.

Analytics has this idea of a funnel for ecommerce sites. People enter a website at the broad rim of the funnel, where there are a lot of places to go. And if all goes well, everyone exits down the same “spout”? that is, the checkout page. Is this analogy good? Is it useful?

There is indeed a funnel pattern, and the way I look at it,all you can do in sales is screw that pattern up. From the user's perspective,the perfect sale is practically a mindless process. You see something you want to buy, and from there, everything else is an obstacle. Even people who love to shop don't love the transaction portion of the shopping experience. Once a person makes the decision that this is the product for them, all you can do is screw the transaction up. The longer the funnel, the more steps the user has to go through, and the more chances are you will eliminate at least some users from the process. That's why Amazon One Click is brilliant.

I find it scary. One click of the mouse and you've committed to the purchase.

It's only scary for those people who have issues with self-restraint. But Amazon has made it very simple and safe?and that's key. One of the messages they put next to their links is that you have 90 minutes to cancel your order. And cancelling is easy?only two clicks.

What about higher up in the “funnel” where people are still deciding whether or not to make a purchase?

It's still the same design goal. Ultimately, this is either a product a customer wants, or doesn't want. In that sense, the outcome is pre-determined. So the design goal is to avoid creating obstacles to that outcome. You do that by knowing what sorts of information people need to make the decision. It might be price, delivery date, description, or features. It may be that all the cool kids have one. So a designer must ask: what is that key piece of information? If you know that and can get it in the process early,everything else is just about closing the deal.

Here's a good example. If you look for digital cameras on Amazon,you'll get list of available products that fit your criteria?just like on many other comparable sites. But Amazon will also tell what day the camera will actually arrive. That's because for many people shopping for a digital camera, the delivery date turns out to be as important as the price?because they may need the camera for a specific reason, like an upcoming event or trip.

Your website mentions Amazon and Zappos as premier examples of American ecommerce sites.

Netflix is another one. There are lots of really good sites out there. There's a site for the big container ships?the ones pirates like to capture these days?that lets you fill a container with thousands of different items, then transport it. This is an exemplary site, but most Web designers'eyes glaze over when you start to talk about the commodities shipping business. Whereas if we talk about Amazon, people get it.

How do the best sites get that way? What are the design teams doing right?

We've identified three fundamental things. The first is that they are constantly looking at what users do and don't do, as well as what users are trying to do. That context will influence how the team tackles the problem. They don't just copy another site. A common misconception is that those other guys must have thought the problem through. But most of them haven't.

The second thing is that really great teams can tell you what the experience of using their design will be like five years from now. If you ask them, they will actually have an answer. I'm not talking here about the design itself, but the experience of using that design. That means starting with the user experience you are providing today. What's that like? Chances are that for many people, it involves a fair amount of frustration. What would happen if you removed all that frustration and substituted a whole bunch of delight? What would that experience look like? You should be able to answer that question.

The third element has to do with the culture of the design organization. Our research makes it clear that you have to have an organizational culture that understands the process of good design, including the role of risk. Creating great designs involves taking risks. Everything that Apple produces isn't perfect. On the same day Apple announced the iPhone they also announced a device called Apple TV, which plays computer-based and online content on a television. When the iPhone was about to come out, people lined up two days in advance to buy one. No one lined up for an Apple TV. The device sells, but it has not had the iPhone's runaway success. Yet Apple TV was designed by the same people with the same goals and methods. So why isn't Apple TV as successful as the iPhone? In a way, it doesn't matter. The important point is that you need to have a culture that will allow you to put both out. You have to have a culture that says that not everything you're going to do will be a success, but we're going to learn from all of it, from everything we do.

Do you conduct research the way people imagine it:  watching people interact with websites?  Do you conduct eye-tracking studies? 

We have done a lot of eye tracking over the years.  It has its place in research, but it doesn't really have its place in Web design: so we don't use it with clients very often.  We prefer either bringing people into a room and watching them work with the design, or going to their site and watching them in their own environment.

Some researchers swear by eye tracking.

Eye tracking has been promoted by a handful of consultants that are just a step above palm readers. The problem with eye trackers is that the results are open to a lot of interpretation. You get a color-coded image that shows in red where the user has gazed for a long time and in blue where the user hasn't gazed at all. Let's say you produce this chart of red and blue blotches for your home page. What does it mean? Do the big red blotches mean that the person looked at it and understood what was going on? Or does it mean that they looked at it and were completely confused?

One of the interesting things about these charts is that scroll bars aren't generally viewed at all, because many users are capable of acquiring the scroll bar through their peripheral vision. Given that, what exactly are these charts measuring? The ambiguity is enough that if I give the same ⁠heat⁠⁠ map to five different eye-tracking experts, I will get five completely different interpretations as to what is happening.

So if it's not eye tracking, what about the analytics idea of watching people's aggregate movements?the so-called footprints in the sand.

That's fine, as long as you understand what the user is trying to accomplish and whether they accomplished it or not. If I have a thousand-click stream showing that users start at the home page and end up on some other page, do I know what those users were actually trying to do? Do I know what percentage of them actually completed that objective? That's the problem.

A great example of this is the Disney site, which we've been testing since 1995. A woman came into our lab who was headed for Disneyworld in Florida. She had a six-year-old who loves trains, so she wanted to book a hotel that was right on the monorail. So here's a reasonable online query: what's the cheapest hotel on the monorail at Walt Disneyworld? I guarantee you that if you try to do that task online, it will take you 20 to 30 minutes. But if you call an agent to make a reservation, they will give you the answer in about 90 seconds. And here's the thing I love: of the people we asked to do this task online?and we've asked hundreds since 1995?about one-in-five will choose a hotel in Disneyland. In other words, they will book a room 3,000 miles away.

Now analytics will tell you that people are indeed booking hotels. But it won't tell you that they are booking a hotel on the wrong side of the country. You won't be able to tell just by the footprints. The important distinction here is between observation and inference. The observation is that someone clicked on the page and viewed it. The inference is that they actually wanted to be there, that they were in the right place. And that's not an inference you can always make. Analytics presents a lot of grey areas like that. When people spend a long time on a page, is that good or bad? When you have a huge number of people clicking through from organic links,but the bounce rate is really high, is that good or bad? The problem with these stats is that they are not telling you the whole story.

The New York Times reported that with some travel sites like airlines, more people are switching from the Internet back to the telephone. You'd think these problems would have been solved by now.

The problem hasn't been solved. Travel is assumed to be a generic process, and it's really not. Booking a vacation with your kids is a completely different process than booking your fifteenth business trip of the year. Even booking a business trip to a city you've been to a hundred times is very different from booking a business trip to a city you've never visited. And these sites don't know how to deal with that.

What's the solution?

I'm a big fan of talking to people. Let me tell you what Netflix does in their DVD rental business. They will take a bunch of design ideas on, say, a new way to watch movie previews. Then they will try them on their website for a couple of days to see what happens. They'll look to see: do people actually watch the preview to the end? Will they take the time to actually rate the preview if given an opportunity? Will they actually add the film to their Netflix queue? Those metrics say something about the relative strengths of the designs. But the design team goes a step further. They bring people into their lab and watch them. First they look to make sure the reactions in the lab reflect what they saw in analytics. Then they talk to people, ask them questions, and get their feedback. And from that, they change the design and repeat the process. They do this very quickly?they push out a new design every two weeks.

Your blog talks about “visual disorientation,” the ability for a designer to zoom back far enough from a site to see it fresh.

Since I wrote that article, hundreds of people have emailed me and said ⁠I do that.⁠⁠ What they do is step back from the design and try to see whether the design they have in their head actually matches what is coming out of their fingers. If it doesn't match, is that because they did it wrong when it came out of their fingers, or is it that their original idea was flawed? You need to correct one or the other. This is something that all good craftsmen do, whether they are building websites, putting in cabinets, or building a boat.

It seems Web designers and developers are taking craftsmanship more seriously. What kinds of changes are you seeing in the profession?

I think all professions, including Web design, create a distinction between craft and art, and it's important to know the difference. If you are building a development of 150 houses, how much craft needs to be involved? Can you just use a handful of proven designs? Or do you need to get Frank Gehry to design each house? At some point, it doesn't pay to make every house a work of art because the costs are too high?and besides, people don't care: they just want a nice house. Every craft has this distinction between satisfying the creative elements and satisfying the basic needs. And the truth is, you first have to meet people's needs. Great designs are wonderful if they meet people's needs and awful if they don't.

A movie that is a great piece of art may touch you so profoundly that you find yourself crying at the end. But websites that make you cry are hardly ever acceptable. Art goes for the whole range of emotions. Art makes you realize something within yourself that you didn't realize before?touching you in a way that is original and distinct. Whereas design is more about achieving a goal. That doesn't mean designs have to be butt-ugly. On the contrary, the aesthetic quality of a design is very important. Apple proved that with the iPod, turning an MP3 player into a fashion statement. But the iPod had to be a great MP3 player too. Just being a piece of jewelry wasn't going to make it work. That's the difference between design and art.

So here's the difference between a great website designer and a great digital artist: the designer looks very closely at both form and function and makes sure that the two work together. That goal is always about eliminating frustration and increasing delight.