Summary: This is the definitive article on user experience, delineating the reasons for implementing UX and the main methods to do so. Show it to bosses or colleagues who must grasp UX. It is also a guide for those aspiring to pursue a UX career.
In the first half of this two-part article about the basics of UX, I explained what user experience is and provided definitions of related terms like user interface and usability. In this concluding part, I explain why UX is not merely advisable but essential and how you should optimally implement a UX process if I have succeeded in persuading you to embrace user experience.
User experience is not some frivolous luxury or decorative bauble. It is an essential discipline that dramatically boosts quality and directly fuels profits. Embrace UX wholeheartedly, infusing its ethos throughout your organization, and your designs will resonate deeply with customers, captivating their hearts and minds.
As a brief refresher from Part 1, in case you can’t be bothered to follow the link, user experience (UX) refers to how people feel about using a product. User interface (UI) is one component of UX that focuses specifically on the on-screen elements. Usability looks at how easy or difficult a product is to use. UX subsumes UI and usability, going beyond surface-level interactions to consider the entire experience of interacting with a product or service.
Why You Must Invest in UX
Following a good UX process has been empirically demonstrated to substantially improve quality metrics — by an average of 75%. This is an average, and some design projects don’t improve much. On the other hand, some projects see incredible quality gains, especially “UX virgins” untouched by prior user research. More than doubling usability after your first study is a common outcome because there is such a thing as low-hanging usability fruit when grappling with a flawed design.
We know that following the UX process can’t produce a worse product because if you test, you’ll see if you have a stinker and can avoid launching that redesign.
Improved UX quality translates to improved business metrics, but how it fattens the bottom line varies by project:
Ecommerce sites have the most evident ROI argument because better design transmutes directly into hard cash through more sales — you can measure that conversion rate blossom. If the customer can’t find the product, the customer can’t buy the product, so findability and discoverability are two noticeable gains for ecommerce sales. Clearer product descriptions and more helpful product photos that answer customers’ main questions are two other common ways user research creates stronger ecommerce sales. Finally, better product descriptions and comparison tables reduce the risk that customers will purchase the wrong product, reducing costly returns processing and increasing the number of satisfied customers who will recommend you and return later to buy more.
Enterprise products, intranets, and other internal products also have a clear economic case for better UX. Since you are paying the users for their time spent using the design, having employees spend less time wrestling their tools equals productivity gains with the associated cost savings. You slash training costs by making new designs easy to learn. Errors are often costly to fix in a business, and good UX work can target the most expensive and common errors and create redesigns that often reduce errors to 10% of their prior frequency. Finally, improved satisfaction with work tools leads to improved job satisfaction and lower employee turnover, which has a financial value that can be quantified if you know your company’s recruiting budget.
Consumer products will often sell more if they are easier to use. An improved product will see more use, whereas a difficult product gathers dust. You’ll also experience better product reviews and word-of-mouth recommendations. Having products people like because they were designed to support users does wonders for branding. Finding an unmet product need in discovery research can lead to true differentiation and the associated branding and sales gains.
Government agencies don’t get direct economic gains from better design but will fulfill their mission better when citizens can find and understand the information on their websites. Also, better UX leads directly to better constituent service, which is the goal of many agencies.
Charities can multiply their donations many times over by more persuasive content design and serve their mission better by improving their website’s findability and understandability.
All organizations, all products, and all services benefit from a branding lift after improving their user experience. Brand is experience in the digital age, and experiencing a clunky product or website bruises brands.
A universal rationale for UX is simply risk management. Your product will be tested by users: your only choice, whether that testing is in the lab before you launch or in the market after you launch. If you test before launch, you reduce the risk of shipping something so terrible that it will severely damage your brand. On a more positive note, you simultaneously increase the chance of improving the product enough that it will enhance your brand.
Finally, the cost of fixing a design weakness discovered by testing during the design phase is about one hundred times less than fixing that flaw if it’s not found until the product is out in the market.
As discussed in the next section, UX work can be pretty cheap, so the cost–benefit ratio of getting the right design is astronomical.
UX is risk management: usability testing before release lowers the risk of flaming out in the market. (Image by Midjourney.)
How to Execute UX
There are myriad models of user-centered design, but the distinctions are trivial compared to the chasm between user-centered and engineering-driven design. Critical elements of any proper UX process include:
Empirical studies of representative users. Don’t rely on what you like or what your colleagues or bosses like. You are too different from the customers for your preferences or personal experience to matter. You know too much about technology in general and your product in particular.
Behavioral research, not opinion research: watch what users do while using products, don’t rely on what people say because that’s often wrong. 62% of all statements are made up on the spot. (As was that bogus statistic, but people do make up the answers to questions on the spot, especially when asked about something they can’t realistically know, such as how they might hypothetically use a new feature. If you want to know, design a prototype of the feature and watch people attempt to use it.)
Early focus on users: before designing a single pixel, start by discovering customers’ needs.
Iterative design: user interface design is so complicated and involves a combinatorial quest through so many design dimensions that nobody can design the perfect user interface on the first attempt. Give it your best try, then test it with users, and return to the old drawing board for another iteration. The more iterations, the better.
Fast and cheap methods are more profitable than slow and expensive methods because fast UX allows for more iterations in the available time.
The “double diamond” model is the most popular approach to UX design. However, I don’t like the single touchpoint between the problem-space diamond and the solution-space diamond, as shown in the visualization below (mea culpa) and in most other illustrations you will find online. This visual implies that the discovery phase has produced a single problem definition as a hand-off point to the design phase.
However, while a clean handoff from problem to solution seems appealing, in practice, the line blurs, as prototyping solutions will reveal new angles to the problem. The double diamond’s linear nature is thus somewhat misleading.
This model correctly captures these dual dualities, giving us sorely needed structure for a messy situation:
There are two spaces to explore: the problem space and the solution space
We explore these spaces by first diverging to consider many options and then converging to narrow down and refine the final choice
Overall, the double diamond is a popular model for a reason.
The “double diamond” model of the UX design process. Visualize time as moving from left to right through the two diamonds, with the height of a diamond at each stage symbolizing the number of alternatives under consideration at that time. In the first diamond, we determine the problem we want to solve; in the second, we design a solution.
Problem Space vs. Solution Space
Resist the temptation to leap into design, especially if you think you know what should be done. Hold your horses and challenge that assumption. It’s often the case that customers need something different than you think — or even what they tell you they want.
A classic example is Microsoft’s customer research in preparation for the design of Microsoft Office 2007. The team assumed users wanted new features and asked key accounts for their requests. Customers described many new desired features, but it turned out that almost all these features were already present in the old release of the product. The real problem was that users could not find their way around the profusion of features in Office.
As a result of this research, the design team pivoted from designing new features to making the existing features more discoverable and introduced the “ribbon” design in Office 2007. This case study is also a great example of why I say that you should not pay attention to what customers say. Customers are experts on their own business but not on the design of your product, so they’ll often request the wrong thing.
The full UX process consists of two parts: The discovery phase, where you investigate the problem space and determine your product direction. Typical methods to use here include field studies and user interviews. We generally take a relatively open-ended ethnography-inspired approach because we don’t know what we will find and don’t want to lock ourselves down too early.
Second, we move into the design phase, where we investigate the solution space to find the best way to solve the problem defined in the first phase. As mentioned, don’t think of the transition between the two stages as a clean handoff with a single point of contact. Usually, when you prototype and test a range of possible solutions with users, you’ll discover additional twists to the problem and refine your understanding of what should be done.
Diverge vs. Converge
Both phases include the same process steps: first, a divergence step where you explore as many alternatives as possible, and then a convergence step where you gradually reduce the options and refine them into a more polished creation.
It’s essential to diverge before you converge because only by considering many alternatives will you sufficiently explore the two spaces (problem space and solution space). Current AI tools can be a great help during divergence because they generate a plethora of creative alternatives for you to consider.
Ideation is free with AI. Ask for 20 ideas, no ask for 50 — doesn’t matter, you’ll get as many as you want in a few seconds. In the old days, design processes and design tools assumed ideas to be expensive, because you had to pay a human designer to sit around and think them up. Therefore, we used to be miserly with redesign requests. Now, if a certain design element fails usability testing, take it out and ask for 10 alternate design ideas. You ask, you get, that’s the new ideation process. Now, the human value-add is how we narrow AI’s many ideas (many of which are inane) into a final product.
To consider many alternatives and get rapid usability findings from having them used by real customers, you should use simple UI prototypes in the beginning, then gradually refine them to higher-fidelity prototypes and final implementation. The longer you wait before committing anything to code, the more flexibility you retain to quickly try out alternate solutions if you discover usability problems. (And there are always usability issues in any early design. If you don’t find any, it proves that you ran your user test wrong, not that you are the one design team in the history of computing to have made everything perfect on the first attempt.)
While divergence is more creative, convergence is more a matter of systematic methodology and judgment, which is currently where humans outshine AI. The critical element here is to employ evaluation methods to drive toward higher quality as you progress through this stage and gradually converge on the final solution that’s released. We need a function to tell us what’s good and bad. Luckily we have such tools in the form of user research. You're gold as long as you decide what’s good or bad based on empirical observation of real customers, not your own opinions.
Research Methods: User Testing is #1
There are many user research methods, and each has its role. But one method rules them all, and that’s qualitative user testing. This is usually called “usability testing” because usability is one of the main quality criteria we test for. However, I often use the phrase “user testing” because the defining characteristic of these tests is that we are subjecting our design to be tested with actual users during real use. Both terms mean the same thing and are often used interchangeably.
User testing can be used at any stage in the UX process. In the diverge step of the discovery phase, you can run a competitive test of a bunch of alternative websites or other available designs. Your competitors’ designs are the best prototypes of your new design because they give you user data about various ways of solving a very similar problem to your own.
In the diverge step of the design phase, you can create simple prototypes of a wide range of design ideas and test them with users. Test early enough, and you can get great insights from prototypes with only 5 screens you have mocked up for each possible design direction.
In the converge step, you gradually build out the prototype and reduce the number of alternatives considered. Proceed through as many iterations as possible because each additional iteration drives up the UX quality. To test many iterations with users, each test has to be fast and cheap, and I recommend testing with 5 users in each round. Yes, you won’t get exhaustive insights from a small 5-user test, but you can run it in a day or two. Then spend another day or two redesigning your prototype to address your identified usability problems. And then test again. If you go through 20 iterations and test each one with 5 users, you will end up with test data from 100 people, but your collected insights will be much richer than if you had tested all 100 users in a single round to beat a single design into submission. And most important, your final deliverable will have much higher UX quality.
Infographic about the benefits of doing iterative user testing with 5 users in each round. Feel free to copy or reuse this infographic, provided you give this URL as the source.
User testing has 3 basic rules:
Recruit representative users. You can’t test with your own employees (unless you’re designing enterprise software for internal use). Test the wrong people, and it doesn’t matter what they do because it won’t translate to customer behavior in the market.
Have them perform representative tasks. Don’t have people play around with the product for no purpose. Don’t just show it to them and ask whether they like it. Your test users must attempt to solve a real problem while you observe what they do.
Shut up and let the users do the talking. Ask participants to think out loud as they perform the test tasks. This allows you to understand how they interpret the design and what they expect to happen if they click a button. Both may be pretty different than the design team intended, and you’ll have to redesign accordingly (either make the system behave the way users expect or change the design to communicate an accurate mental model better). Avoid biasing users with comments or suggestions: sit quietly and watch what they do independently.
Without empirical user data, you’re designing blind. Programmer intuition is no substitute for observation and insight from real customers. Their priorities differ drastically from your assumptions. As shown by my three bullets, it’s so easy to get real data. Don’t miss out.
Recommended UX Process in 8 Bullet Points and Two Haikus
Discovery first, design second.
Divergence fosters creativity.
Convergence drives quality.
Converge through user testing.
Iterate, iterate, iterate.
Base each step on customer insights.
Prioritize qualitative insights over quantitative data.
Prioritize speed over fidelity.
A more poetic way of summarizing these points:
Double diamond's path, Diverge, converge, iterate, Design's dance unfolds.
Usability, Low-hanging fruit to be plucked, Profits blossom bright.
Can’t We Just Hire a Good Designer?
Some people don’t believe my poems and dismiss the need for even my 8 simple steps, claiming that it’s enough to hire a good designer. He or she will create an excellent design without needing discovery research, user testing, or other extraneous steps. It’s more or less the definition of “good designers” that their design work is good.
First, let me point out that if you’re a company with this attitude, you won’t have great designers working for you. The very best designers are motivated by creating outstanding designs that push their talents to the next level, and world-class designers build upon user insights and iterate relentlessly.
In general, the laws of statistics dictate that most companies have average designers, and many companies even have below-average designers. So even if you could get one of the world’s top designers to do good design with no overhead, that’s not whom you will usually have available to work on your project.
If you are among the few lucky companies with world-class design talent on staff, you will find that they create better designs than their less-talented colleagues elsewhere. But even these great designers will produce even-better work if they (a) create initial draft designs based on an understanding of users founded in research and (b) iterate the designs through subsequent versions, polishing each version after getting feedback from usability testing.
Conclusion: UX = Profits
After 5,000 words about user experience (across parts 1 and 2 of this “UX What Why How” article series), what's the bottom line? I could spill another million words about twists, turns, special circumstances, and advanced methods. I have even done so in my many other books and articles. But the most important basics are what I just told you.
Make user obsession your north star. This mindset shift separates those reaching greatness from those languishing in mediocrity. Your UX quality will increase dramatically, leading to more sales, higher customer satisfaction, and a better brand reputation. All of these equate to more profit, which is my one-word conclusion. UX is cheap compared to what it’ll gain your company.
Your customer: what is she doing, what is she thinking, what does she want and need, what are her behaviors? Do UX right, and you’ll know the answers to these questions, creating a firm foundation for profitable design based on a full-color picture of the user. (“Computer user” generated by Midjourney.)
Infographic to Summarize This Article
Feel free to copy or reuse this infographic, provided you give this URL as the source.