An interview with Peter Merholz and Nathan Shedroff on User-Centered Design
Digital Web: What is user-centered design (UCD)? What are its benefits? Problems?
Peter Merholz: Contrary to common wisdom, user-centered design is not a process, but a philosophy. User-centered design requires the inclusion of a product’s end-users throughout the design process.
The primary benefit of user-centered design is that, when performed well, it ensures that the product is useful, usable, and meaningful to the end-user. Also, many of the low-fidelity methods developed to accomplish user-centered design allow for shortened development cycles.
The primary problem of user-centered design is that people engage in it at the expense of all else. Oftentimes, what is *most* useful, usable, and meaningful to the end-user is untenable from a business perspective, and the product, while maybe popular, is a financial failure. Additionally, UCD can often get bogged down in process, in needing to verify every design choice with users, unnecessarily encumbering progress.
Nate Shedroff: User-centered Design is an approach (with many variations) to creating experiences (whether products, interfaces, events, or other media) for people with their needs in mind. Usability is one of the primary focii but only one of several. Others include usefulness, desirability, legibility, learnability, etc. The benefits are that these experiences are often easier to use and learn; more appropriate in terms of functions and use, and more compatible with existing processes.
There are several possible problems with this approach but they are few compared to not using a user-centered approach. One problem is that the users targeted for the experience are either too narrow or not reflective of the true users (in personality, needs, uses, behavior, experience, etc.). Often, the choosing and defining of “users” in more political than it should be and less reflective of reality. Another possible problem is that, despite using a user-centered model, developers ignore important issues and feedback as they discount things they don’t really want to address. This happens all the time.
User-centered design shouldn’t be confused with participatory design which is an approach that requires actual representative users to participate in the designing and project specifying processes. This can have success over a wide-ranging degree. Sometimes, user involvement to this level becomes political and intrusive. Other times, users only know to ask for things that they’ve already seen and there is no mechanism for truly innovative solutions. However, it is always worthwhile to test solutions with users in order to get feedback during the design process when changes and enhancements can still be made.
Digital Web: Usability. User-centered Design. User Experience. User-this-n-that. What’s the difference among these terms?
PM: Usability: the quality of a product that makes it usable (i.e., understandable, workable). User-centered design: a philosophical approach to design that relies on the inclusion of end-users throughout the design process. User experience = the totality of an end-user’s experience with a product, including its usability, but also its aesthetic appeal, functionality, desirability, and other attributes.
NS: Usability is specifically focused on “ease of use” and is a component of User-centered Design. User Experience can mean a lot of things depending on who you’re speaking with but it should be the focus of a user-centered design process. The experience encompasses a lot of space, including 3 dimensions, 5 or more senses, time, emotions, social context, etc.
Digital Web: What is a usability analyst’s role?
PM: To assess the usability of a product or system. Typically, they conduct tests of products with users, running them through tasks, and from a set of tests, are able to highlight a range of problems with the product or system, and suggest recommendations for improving the usability.
NS: I believe it means someone who specifically analyzes an interface or experience in terms of usability (easy of use) but that seems way too specific to be a whole role for two reasons. First, usability is just one aspect of User-centered Design. Second, if all someone does is analyze and never develop as well, they can’t do a very good job of analysis. Design is an organic process, as is research. If you aren’t doing a bit of both, chances are you’re not doing either well. You find things out in the process of design that would never come-up in straight research and visa versa. Both inform the other. If you spent all your time on one and not the other, you aren’t being well-rounded enough to do a good job.
Digital Web: Navigation falls under IA, but it has fallen under User Experience. Is there a difference?
PM: I think of information architecture as a subset of User Experience, and, as such, navigation, which I typically consider as part of information architecture, thus also falls under user experience.
Navigation and information architecture are tightly related, though distinct. Think about a building. A building’s architecture refers to its structure, the placement of systems (wiring, plumbing, etc.), the placement and size of rooms, the means by which people will traverse space, etc. A building’s navigation (typically called “wayfinding”) is a system of signs and cues as to how to move within the space, to find the desired destination. The navigation is reliant on the architecture, but if the building were just rooms with unmarked doors, no one would know where anything is. The navigation adds a layer of information over the architecture, directing visitors to their desired goals.
NS: There’s a lot of disagreement here, but User Experience is a larger umbrella and IA (and navigation) is just one of many items under that umbrella. Navigation is one component of the information design (or architecture) but certainly not the only one. IA is one of several disciplines that make-up the whole experience.
Digital Web: What heuristic evaluation? What are its strengths and weaknesses?
PM: Heuristic evaluation is a usability engineering method developed by Jakob Nielsen and Rolf Molich in the late 1980s. Jakob and Rolf came up with a list of ten heuristics for successful interactive design—such as “Visibility of System Status” and “Recognition Rather than Recall”—which can be used by an expert evaluator to assess a system, and make sure it’s usable.
Heuristic evaluations are great for performing usability on a tight deadline—they can be completed in a couple of days, with little up-front preparation. Also, unlike user testing, which by nature responds to the tactical aspects of an interface (since users can only really respond to what’s in front of them), these evaluations allow experts to make strategic recommendations based on what they’ve seen.
There are two primary drawbacks to heuristic evaluations. The first is that, really, there is no substitute to talking to and observing users. A trained expert can get a lot of the way there, but users have a habit of being delightfully unpredictable, and you won’t know for sure of your site’s usability without watching them. Second is the tendency to treat heuristics not as “rules of thumb” but simply as “rules.” Like any guideline, these should be understood as what they are—aids not strictures.
NS: A heuristic evaluation is a valuable technique but one of many. It really depends on the project and process as to its effectiveness. Its strengths are that it makes specific use of known processes or functions and it is a very easy method to employ for an analysis.
Its weaknesses include only relating to known problem types and not being able to find problems of unknown function. It assumes that everything about how good and bad interfaces and experiences work is already known and documented (and in the evaluation criteria). However, we’re always learning more about what we don’t know and we have a lot to learn and add to these criteria concerning the sensorial, spatial, emotional, and social aspects of experiences. Simply employing a heuristic evaluation is, by no means, a complete evaluation of an experience.
Heuristic Evaluations also usually work best on simpler experiences since there are usually fewer dynamics to discover. The dynamics of complex experiences can grow exponentially.
Digital Web: What other inspection methods should be considered? Why?
PM: It all depends on what you’re inspecting. And, it’s better to “inspect” midway through a design process, rather than waiting until the end. Typically, usability doesn’t come into play until the design has been largely settled upon, at which point it can only serve to tweak the design.
Ideally, users will be incorporated into the design process early on. As designs are first being considered, you should have your prototypes submitted to “guerrilla user testing,” a low-key, fast-result method where you bring 4 or 5 people in on a day, run them through the designs, see if they could accomplish the desired tasks, and then, the next day, roll what you’ve found right back into your design process.
With Web sites, a huge concern is accurate and meaningful labeling—since it’s words that your users are clicking on, those words better make sense. You can test your nomenclature by showing people screens and asking them what they think they’d see if they were to click on a particular link or button.
Additionally, Web sites typically require marshalling vast amounts of information, and it’s a challenge to create useful and sensible categories that work for the majority of your users. You can “user test” your categories through card-sorting.
First, your design team comes up with categories for the site’s content. You can test those categories in two ways—in a closed sort: you give users the required categories, and then give them 40-50 cards representing pieces of content, and have them organize them as they see fit.
In an open sort, you simply give them the 40-50 cards, have them create the categories themselves, and then label them. A closed sort will lead to a more rigorous and statistically significant result, but could limit you from discovering better categories. The open sort will offer a better view into your users’ heads, but the variety of results might be too great to sensibly manage.
NS: User Discovery in its many different forms is always a good idea—especially with real users in their real contexts. There are too many things that we take for granted in the use and design of experiences that can only be discovered by observation.
Digital Web: What’s the connection between usability and accessibility?
PM: I would consider, and this might get me in trouble, that accessibility is an aspect of usability. If someone can’t physically interact with the site, then it’s not very usable, is it?
NS: Well accessibility has two meanings. The more common is an overview of the myriad forms of disabilities (such as vision loss, hearing loss, motor impairment, etc.). The other definition relates to the ability of people to have access to specific technologies (such as computers or the Internet). Since the first definition is more common, I’ll address it.
Usability relates directly to this definition of accessibility since Accessibility is really Usability that includes these disabilities. This means that Usability and Accessibility are mostly the same thing except that Usability usually assumes that users don’t have physical, mental, or emotional disabilities (since the majority of most of the public don’t possess these). Widening the user requirements to include those with these conditions simply widens the usability requirements and needs—specifically the user testing.
Business of UCD
Digital Web: How does UCD play a role in ROI?
PM: I can write a book on this. Some have. To put it simply, I see UCD as playing two essential roles in ROI.
The first is in increased revenue—by practicing user-centered design, you’re more likely to develop products that people are interested in using, and more likely to develop products that people are capable of using, both of which should lead to greater use of those products and, concomitantly, increased revenue from sales or subscriptions.
The second is decreased costs, which in turn gets divided two ways. The typical way is to monetize the gains earned through efficiency by creating something that’s easier to use—if it takes people less time to accomplish a task then that costs less money. Less commonly recognized is how UCD leads to operational efficiencies—by following a UCD practice, your development cycles are shorter, the resources on your team are more wisely used, and the products you’re putting out are better, not requiring as many further versions to “get it right.”
NS: It can have a huge affect but, for the most part, is never measured or addressed. For starters, the usability of an experience or interface often has an effect on the quality of work. Therefore, for important—and critical—jobs, the usability can drastically influence the number of errors made, the amount of work produced, and the quality of that work. It can also change the conditions of that work—particularly the stress level of users.
Second, usability can change a user’s interpretation of the product or company’s brand. At the time of purchase, a more usable or obvious interface may make a difference in the sale of that product. Certainly, it can affect a user’s enjoyment and loyalty toward that product and company for future purchases and advice to others.
Usability can also affect the amount of support, maintenance, and customer service. These aspects of business are rarely linked to the design process of products and services yet better solutions can often drastically cut-down the needs for these corrective measures, directly saving a lot of money.
Digital Web: How do you optimize UCD with tools and technology?
PM: Very little. UCD is an approach that can be done with pencils and paper as well as it can be done with computers and electronic whiteboards. I leave it up to individual practitioners to utilize the tools that best suit them.
NS: I imagine that well-developed and appropriate tools can ease the management of the development process and the collection of data and solutions, but they can just as easily hinder and destroy the process. Too many tools and technologies already on the market make it much more difficult to develop products, services, and experiences from a UCD perspective by being cumbersome and unwieldy.
Digital Web: What are the ethical issues that can come up when working in usability?
PM: I can’t say I’ve ever faced a usability-related ethical issue. Making products and systems more usable is pretty ethically sound, no matter how you look at it.
NS: I imagine that the most pressing one is that, sometimes, companies don’t want their products to be very easy to use or understand because they’re counting on selling service contracts to their users that are much more involved (and, therefore, lucrative) if their products are difficult to use and maintain.
Another common issue (which isn’t so much ethics as merely priorities) is that developers make decisions many times against what their research tells them is the right thing to do (or ignore clear problems) simply because they want to do something they consider “cooler,” more beautiful, or simply different based on their own ideas and desires. This is why we have so many Flash intro movies on Web sites, for example. However, this isn’t so much about ethics as priorities.
Digital Web: How can Web design shops “sell” usability to clients who say things like, “What know what the users want.” “Our budget is too tight for that.” or “It was easy to use.”?
PM: “Selling” usability to a client is usually a non-starter. Clients aren’t interested in usability; they’re interested in results, in meeting objectives. If clients feel that they “know what users want” or that something is “easy to use” then, no, you’re not going to convince them otherwise.
Though, those clients must be coming to you for a reason. Dig into that. Find out what’s not working for them. If they say that “customers aren’t satisfied,” or that “only our developers know how this works” or that “the percentage of people who register is low,” those are all triggers that the problem might be a usability one. At that point, say, “We have some methods that might answer your problem,” and talk to them about usability.
For Adaptive Path, our problem isn’t selling usability—it’s dissuading clients from buying usability. Oftentimes “usability” is the only thing they’ve heard, and so they want you to do that. However, when we talk to them, we often discover that the problem isn’t “usability,” but typically one of poor development processes that didn’t involve users from the beginning of design.
NS: Well, there’s never any way to truly know how usable an interface is without testing it. However, this is always the first thing to go when trying to cut-down a schedule or budget. Developers need to impress upon their clients that the ROI for more usable experiences can easily offset the costs of testing and other usability techniques. However, the problem usually lies in the fact that the effects of these changes almost always lie in other divisions of a company (such as marketing or customer service and, therefore, aren’t a real concern to the division purchasing the development work (since it doesn’t impact their bottom line). The merest amount of user testing, for example, can be invaluable and plainly show problems in a design. Even simple paper prototyping within a development house with other employees (not working on the project) can be used to improve the design and sometimes convince the client of the need for more extensive research. For financial, security, and service-oriented services (such as financial management sites), it is imperative that the interface work well for the majority of users. The consequences could be catastrophic.
Digital Web: Measurements – they’re a nightmare for many. Measuring the user experience seems impossible, but it can be done and it should be done. What is the process of measuring it? What should be measured?
PM: You’re right that it should be measured. My friend, Alistair Williamson, has an entire company (WebCriteria) built on measuring user experience. However, user experience can’t really be measured in the abstract. Alistair’s approach, which is as good a one as I’ve heard, is to acknowledge that most sites have tasks which they want their users to accomplish. And that there are beginnings, middles, and ends to those tasks that you can follow.
If people aren’t able to accomplish a task, it’s likely because there’s a problem in the user experience. And if you alter the experience, and more people accomplish that task, then the design is successful. What you specifically measure, though, is dependent on the business, on their objectives. Things like “page views” and “conversion rates” are meaningless without context.
NS: There are too many processes and techniques to mention and these should include many of the techniques market researchers use to evaluate customer reactions to possible solutions.
Digital Web: Usability experts are encouraging Web design shops to take a user-centered approach to design. What were they doing otherwise? Isn’t it always about the audience?
PM: Ha ha! I’ve heard this one a number of times. And the answer is: no. In my experience, Web design is primarily about the business. They have things they need to get out there, and they will, whether or not any of their customers care.
Secondarily, Web design is about the designer. They take the business requirements, and create designs that satisfy themselves, and, they hope, win them an award from Communication Arts magazine. While the business and the designer will often talk about “the user,” it’s rare that they have a firm understanding of their audience.
It’s this simple: user-centered approaches require that you engage the user in the design process. Many Web design shops were simply not doing this.
NS: You can’t draw that conclusion if you look at the majority of Web sites on the Internet. Too many designers are trained to design for themselves. That’s how most design programs work. Engineers are trained similarly and are usually never asked to consider the user. Most designers and engineers aren’t ever asked to work in a group until they get their first job in the industry and many have no skills or interests in doing so.
Lastly, marketers are taught that they have some kind of superior processes and ability to know the requirements of the final solution and this usually blinds them to both the courage and the possibility to truly innovate.
Digital Web: How would you describe, at a high level, the user-centered design process?
PM: A fairly standard model has emerged over the last 5 or 6 years. It begins with some form of user research, whether it’s interviews or surveys or contextual inquiry. That research is then used to define the problem at hand, creating models that eliminate the muddiness and attempt to make the audiences’ needs clear. Those models lead to the design of a solution that meets those users’ needs.
Digital Web: Hottest thing today is creating personas. How would you go about doing this for a site like Amazon, which has many different types of users?
PM: Well, the first thing I’d do is question the assumption that we’d want to create personas. Amazon’s biggest challenge is most likely one of information architecture—to get people to products that they are interested in buying. Personas are good for supporting interaction design—determining features and functionality and their relationship to one another. Personas aren’t so valuable for information retrieval tasks—the strategies for uncovering the right item are simply too varied to be well-represented through personas.
That said, personas will be useful for designing features that support the shopping task. Because of Amazon’s massive offering, you need to approach the problem piecemeal. I would focus on a particular part of the site, say, the electronics store, and talk to the chief merchandiser about how they segment their customers. Those segments would drive user research, where I’d talk to 6 to 8 folks who fit those segments about how they purchase electronics goods. I would then break down that user data into a collection of discrete elements, demographic, Webographic, procedures, etc., and then build those back up again into two or three personas of typical shoppers for electronics goods.
NS: You need to create a lot of personas. Sit down with the marketing department and review the depth and breadth of current customers, as well as all of the users of the site (including partners, suppliers, vendors, third-parties, etc.). Then, you need to discuss the topic of possible new customers and users with marketers and help develop reasonable and accurate personas.
Next, focus on a reasonable number of personas, though for a site like Amazon, this still might be many. Developers also need to be realistic about the descriptions and parameters of personas. If accessibility issues are important, these differences need to be built into some of the personas. Developers also need to describe some number of scenarios and personas in which people just don’t “get it”—where users truly do stupid things since that’s the reality of the world. I’ve never actually seen this done, but it’s critical for consumer sites.
Digital Web: How do you recruit users for a usability test? After all, users are customers. How do you bug customers to use a Web site to see how they use it?
PM: This is actually very straightforward. I recruit subjects through a firm that specializes in finding people for market research. You give them a sense of the demographics and other qualities that you’re looking for in your users, and they go out and find those people for you. It’s remarkably easy.
NS: You need to carefully select just the type of users you targeted in the development phase (the personas you designed to).
Digital Web: A writer mentioned that when discussing UCD, it is rarely done from a design or creative perspective. What does creativity mean to you? Does it tie in with UCD?
PM: That’s because that writer is either foolish, ignorant, or mixed up “user-centered design” with usability. When used appropriately, UCD drives inspiration—how better to come up with brilliant product ideas than studying the people who will use them and figuring out how to better serve them?
NS: Well using a UCD approach doesn’t imply that everything is already known and quantified and the process plays-out like a subroutine. The reality is that any process is just a scaffold, a structure that can add purpose, direction, and organization but can never cover every possible event and outcome.
There is a tremendous amount of sheer creativity involved in solving problems. The process itself doesn’t create solutions. It is only there to help structure the flow of work to ensure that important issues are addressed in an order that helps save time, energy, money, and work. The idea is that certain questions shouldn’t be addressed until others are asked and answered since they rely on these answers as part of their solution. No one knows where the best ideas are going to come from. It could be anyone on the team—or someone else who just happens by at the right time.
A good process is there to ensure that everyone is, at least, focused on the critical issues at any given moment so things don’t get forgotten and unaddressed until it’s too late. This is one of the reasons why multi-disciplinary teams are so important—from the start of the project and throughout its development cycle. Groups who only bring in specialists, designers, engineers, etc. at the moment in the process when their decisions need to be made run dangerous risks of not finding the best solutions—or potential problems—before it’s too late.
Digital Web: Who has inspired you? Why?
PM: I have been inspired by a range of folks, too many to list, but here are some off the top of my head:
Nick Ragouzis (www.enosis.com), for encouraging me to continually challenge my conceptions. Betsy Martens, for brilliant flashes of insight. Peter Morville (www.semanticstudios.com), for making connections. Karen Holtzblatt (www.incent.com), for clearly bridging the research and design gap. Donald Norman (www.jnd.org), for practically starting it all. Scott McCloud (www.scottmccloud.com), for his cleverness and fervency. Bill Viola (www.billviola.com), for creating transformative experiences
NS: There is inspiration all around us. Specifically, some of my inspirations are great, innovative people such as Tibor Kalman, Charles and Ray Eames, Phillipe Starck, Arnold Wasserman, Niels Diffrient, Richard Saul Wurman, and Brenda Laurel.
Performance artists such as Laurie Anderson or theater designers such as Julie Taymor are great designers to watch. There are other designers in every profession, artists, writers, scholars, etc. to be inspired by. A book like Mike Ridley’s Genome, for example, is beautifully written and organized. Information design is just as powerful and important to the reading experience as the test itself.
Digital Web: What do you see next for UCD?
PM: Next? Waking up to business. At the last Advance for Design Summit, it became clear that the next step for experience design is to turn its gaze away from its navel, to look outward to those whom user-centered design supports, and who support user-centered design, and begin a dialogue to help them understand the value we can bring. For many, this will push them out of their “comfort zone,” but it’s a necessary step for the evolution of the field.
NS: Well, we have a lot of work yet to do. We already have a fine body of work that simply isn’t being employed enough and in enough places. For example, name one interface for a VCR or digital camera you’re seen that you’re happy with. There will always be more to develop and discover and we drastically need to discover techniques to evaluate the social and emotional aspects of interfaces (some of the “next” things for UCD) but we already have made a lot of progress that isn’t having the impact it should. We need to work on getting these tools and understandings into more corporations and development groups.