Summary: Diminishing returns for UX signal victory: usability battles have been won. Users manage, interfaces suffice. AI will juice UX ROI briefly, but don’t overinvest. User experience is no longer special but has become normal, so expect long-term alignment with standard business metrics. User research will sustain usability improvements, but don’t expect the ROI miracles of the past.
The UX design lifecycle, with its many methods, has been getting lower return on investment (ROI) in recent years. While this may seem unfortunate for UX, it’s actually good because the lower ROI is a consequence of our hard-won victory:
UX design has been steadily improving over the last 50 years and is now reasonably good. The low-hanging fruit of terrible usability has long since been picked.
UX has become more established, with well-documented best practices and design patterns available for the picking. This means that new designs are almost guaranteed to have acceptable usability as long as the designers can resist the temptation to violate Jakob’s Law and invent new, weird interaction styles.
The user population is saturated in rich countries: almost everybody with sufficient IQ to perform knowledge work is already using computers. Most people now learn computer use as children, when they are unpaid and spend their time learning many other things as well.
The last point is the least important, but still worth considering when comparing present-day UX ROI with that of the 1990s: Back then, highly paid business professionals were starting to use computers for the first time, resulting in substantial training costs that could be reduced with better usability. Children absolutely face usability problems when learning computers, and usability can (and should) improve further, but nobody is paying for the time kids waste picking up computer skills.
The proverbial low-hanging fruit: Usability was originally so terrible that the fruit wasn’t even low-hanging, it was already on the ground, and all we had to do was to pick it up. Those happy days for UX professionals are past, replaced by a world with decent design. Now, only high-hanging fruit remains to be plucked from the UX tree. (Midjourney)
Let us consider three ages of user experience work:
The truly bad old days before personal computing. Roughly from the invention of stored-program computers in 1945 until PC became common in the United States around the year 1995. (The IBM PC was launched in 1981 and the Apple II dates to 1977, but these devices were originally used by a small minority of enthusiasts. Even as late as 1995, only 34% of U.S. households owned a personal computer, with lower penetration in most other countries. Even so, for simplicity’s sake, I’ll count 1995 as the year computers had become mainstream.)
The Web explosion: roughly 1995 to 2020. Those were the glory days for UX ROI, especially in the beginning.
UX Saturation: roughly starting in 2020. By 2020, major design projects had already become pretty good based on several past years of UX work, meaning that future design advances have become expensive and likely to yield less benefit to the company. Certainly, with AI, we are at the birth of a new interaction paradigm, but the likely outcome is declining importance of user interface design as users turn to AI agents instead of using UIs designed by content and service providers.
The early years of UX were marked by a high return on investment (ROI). Over the last twenty years, the factors that contribute to UX ROI have turned against those high returns.
The Bad Days of Terrible UX: 1945 to 1995
In the pre-PC days, computer usability was abhorrent. Young people today don’t know how good they have it, compared to when I was a computer science student in 1976 and was expected to use a mainframe with a line-mode UI and documentation written by a bunch of programmers who had never taken a tech writing course. (The bad UI of my student days is why I changed from CS to usability, so it did a little good.)
With terrible usability, you’d expect high ROI from making things better, but this mostly didn’t happen. In the pre-PC days, computers were exactly not personal devices, but were only used for specific business tasks for which the operators received extensive training. Yes, I always argued that the cost of training courses is the price of bad usability, but when employees spent most of their working days on a handful of screens, it certainly could (and was) argued that it was better to train people to a high level of skill with those screens than to engage in the mysteries of user-centered design which was hardly practiced.
(When I graduated college in 1983, there were three jobs in Denmark with the mission to make software easy to use. The UX design discipline existed, but most companies didn’t know about it.
Because it was very expensive to develop software and because users were forced (and trained) to use whatever design the programmers devised, there was little incentive to employ UX design teams and utilize iterative design and other practices that later became the standard for good design.
There were, of course, a few exceptions to the rule that it didn’t pay for computer companies to worry about UX. There were those three jobs in Denmark, and a few companies in the United States and elsewhere were building early UX teams. IBM had several groups, including a central User Interface Institute at its research headquarters, where I briefly worked. And the phone company had a well-funded effort where I worked for several years.
Those were big companies with so many users that they found it worth investing a little in a few usability improvements. (For example, some of my phone company colleagues generated $7 million per year in savings from a redesign of the directory assistance software that made it one second faster for an operator to look up a phone number. A small usability gain per transaction summed to big bucks because of the many calls to directory assistance back then.)
There are many reasons UX was neglected before the Web revolution.
The Web Explosion: 1995 to 2020
In 1998, I charged $10,000 for half a day of usability consulting: I did a quick review of the client’s website, followed by a one-hour conference call with the product team (those were the days before video calls) to tell them the top 10 things they should change on their website. Implementing that consulting advice usually required very little because my main recommendations were usually to remove stupid design elements that repelled users and to change the text on the homepage and other key locations to better explain what the site was doing.
Based on this half-day of usability consulting, the client would usually gain at least a million dollars in Internet business, or 100x what they paid me. A rough ROI calculation comes to 10,000%.
Many websites were built with little regard for usability — complex navigation, confusing workflows, and non-standard interfaces were common. As a result, even basic UX fixes led to huge jumps in performance, with a doubling of key performance indicators (KPI) being common. Early web design lacked consistent patterns, so introducing now-familiar elements (clear navigation bars, shopping cart icons, straightforward checkout processes) often boosted conversion and sales dramatically.
In practice, this meant that modest spending on user research or design could double or triple conversion rates for a site with poor UX. Cheap user research and big sales increases created UX ROI in the hundreds or thousands of percent in the 1990s.
Early websites were so bad that even the smallest usability effort would substantially increase earnings, leading to a stupendous return on investment (ROI) for UX. (Leonardo)
Diminishing returns from UX efforts are real. Early on, fixing obvious UX problems (broken links, unclear labels, non-responsive buttons) yields huge benefits; but once those are fixed, further refinements produce smaller incremental gains.
Come the 2010s, most websites and apps followed established conventions (e.g. standard checkout flows and familiar icons), so users rarely encountered completely unusable designs anymore. Improvements became about optimizing details — for example, reducing a form from 5 steps to 3 might increase conversion a bit, but not nearly as much as the initial jump from a clunky 12-step process down to 5 steps.
Though ROI was lower by 2010, it was still high because the amount of money that was generated through user interfaces had exploded. Instead of the few mainframe users of the 1980s, most people were using the Web in 2010 (at least in the rich countries that matter the most for the economics of UX), and they were spending increasing amounts of money online (or having their spending decisions influenced by online content).
There was also a short-lived renaissance of low-hanging usability fruit in the first years of smartphones, starting in 2007. History repeated itself, with the first mobile designs having low usability because they were ported from desktop websites instead of being optimized for the small screen. However, history only rhymes and doesn’t repeat identically. This time around, many digital properties had UX teams and analytics efforts and fairly quickly realized their mistake.
Most famously, Facebook made a major pivot to mobile usability in 2012, despite having spent around 80% of its UX resources on desktop design up to that point. However, as soon as they made that change, they saw a 20% increase in user time spent on their mobile app compared to the desktop product. (With the associated monetary gains from increased advertising income.)
There used to be a wide gap between user needs and what computers supplied. UX was the solution to bridging this gap, which is why it has high ROI around Year 2000. (Midjourney)
UX Saturation = Low UX ROI: Starting 2020
UX saturation is a state where a company has been investing in UX efforts for several years, meaning that all the low-hanging fruit has already been picked. Happily, this means that usability has become decent, if by no means great in most cases. However, further advances will be challenging to achieve and require extensive research and intensive design explorations. It’s not obvious how to improve the UX further, once this state has been reached. Worse, design improvements are now very expensive because of the amount of user research and exploratory design work needed to advance. Worst of all, since the big usability improvements have already been had, those additional design changes will likely yield fairly modest gains in the business metrics that are the goal of all design.
In website optimization, for example, once the biggest problems are solved, you may need dozens of A/B tests (each costing time and resources) to find a few that move the needle. One CRO calculation showed that if only ~1 in 8 tests is a big winner, a team might spend 16 weeks and $60k running experiments just to find one significant improvement. That improvement might be, say, a 2% increase in sales — still valuable, but when weighed against the cost of finding it, the ROI percentage is far lower than earlier “one and done” UX fixes.
Consider user research costs. In the 1990s, a simple hallway usability test or a handful of user interviews (costing a few thousand dollars) could uncover a flaw that, when fixed, doubled a product’s adoption. Today, companies will often need extensive UX research, including multi-week ethnographic studies, multiple rounds of usability testing, and in–depth analytics, to understand subtle user preferences. These research efforts can be expensive, involving specialized researchers, tools, and participant recruiting.
The incremental improvement from applying the findings might be relatively small (e.g. a few points increase in task success rate or a slight boost in customer satisfaction), especially compared to the “quick wins” of yesteryear. The cost-benefit ratio has tightened: UX teams must justify their value by aggregating many small improvements or by preventing costly missteps, rather than by pointing to one big post-redesign spike in metrics.
Furthermore, as UX has been embraced, team sizes and budgets have grown. Large tech firms now have dozens or hundreds of UX professionals (designers, researchers, content strategists, analysts), including several levels of progressively more expensive managers and design executives. (UX “pancaking” will hopefully rectify this management overhead, but for now, it’s real.) This makes UX a substantial cost center. When a company with a mature product invests, for example, $1M/year in UX personnel, achieving a 5% conversion rate increase on an already optimized platform might or might not clearly net a positive ROI in immediate revenue terms.
In the 2000s, spending $50k on a redesign that yielded a 50% conversion bump was a no-brainer ROI. Now we spend an order of magnitude more to get an order of magnitude smaller uplift. This is why, in recent years, some executives have scrutinized UX research costs, especially if they don’t see a clear short-term payoff.
During economic downturns (like the “UX Angst” I declared in 2023–2024), many companies downsized their UX research teams, citing the need to “focus on profits” and questioning whether such extensive research was truly necessary. This doesn’t mean UX research lacks value; it simply means the value is lower than before. The ROI calculus for UX has shifted: it’s less about massive gains from a single project and more about sustaining improvements and preventing losses over time.
Higher cost and lower benefit = lower ROI.
Certainly, a low ROI doesn’t mean zero or negative ROI in most companies. Very few companies have invested in serious UX efforts long enough to have reached the point where additional usability advances will cost more than the change in business metrics is worth.
However, the argument for further investments in UX is weak at this stage. UX efforts should probably be kept in most companies, partly because other companies are still improving their design and thus raising the bar that must be met to retain customers. However, substantial additional investments are not required.
Additionally, UX saturation implies the death of most UX and usability consulting agencies. You can only overcharge for UX advice when it’s rare. Commodity UX is great for users but ruins the consulting business model.
How can I argue that UX saturation implies companies should not expand their UX teams while also expecting the UX workforce to grow by 5x by 2040? Squaring this circle requires the realization that saturation has only been achieved at companies that have already invested heavily in UX for ten years or more.
Most companies either still have nascent UX efforts or are not doing any UX work at all. These companies still benefit from the establishment of industry norms and design patterns which they can copy for free. But knowledge of the specific pain points of their particular customers will require user research, and designing to solve these pain points, once identified, will require all the usual steps of the UX design lifecycle.
Thus, I expect UX to grow in new companies, but not in companies with long-established UX teams. Furthermore, most of the growth will be in countries without the long UX tradition of North America and Western Europe. As a simple example, the highest growth rates for people subscribing to my email newsletter over the last year were in countries like Poland, which grew at twice the worldwide average.
A final point about UX saturation is that, although I dated it as starting around 2020, it is actually an individual event happening one company at a time. Some companies reached UX saturation in the 2010s, while others haven’t yet reached this point.
Will AI Goose UX ROI?
The AI revolution will increase UX ROI, but probably only temporarily, for two reasons:
The cost of UX will decline dramatically because of the huge productivity gains from AI in both research and design. It’s likely that by 2040, 75% of the UX work that used to be done by expensive humans will be done by AI. (The reason this won’t cause unemployment is that much more UX work will be done, once it finally cheap.)
AI is unknown land for UX, and we’re still in the phase where we are establishing the best practices for AI products and AI-infused features. This again means that big gains can be had for those companies that invest in user research for AI to find out how people use these new capabilities in their work and everyday life. However, this landgrab period will likely be short, just as it was for mobile, because of the established history of e-commerce and mobile. Design patterns will be documented soon enough, particularly since AI doesn’t require nearly as much UI design as legacy software.
Lower cost, higher gains = high ROI, at least for a few years. However, the long-term trend will assert itself, and UX ROI will decline to the same levels seen by other disciplines.
In Iceland, the Landgrab Book (Landnámabók) documents the settlement of the island by Vikings in the 9th Century. In particular, the book identifies the 435 individuals who were the initial settlers and claimed the best land for their homesteads. Arriving in a new territory early was valuable then, and still is. (Midjourney)
ROI Trends
The following chart shows my rough estimate of the ROI for UX spending over a 60-year period: low numbers during the tail end of the mainframe era, slowly rising ROI during the PC era, a huge spike during the dot-com bubble, followed by strong but slowly declining ROI until the present day. I predict a slight increase during the initial phase of the AI era, followed by a continued decline until UX has a return on investment (ROI) comparable to that of any other business activity. (Even though this looks like zero in the plot, the future expected ROI for UX is probably around 20% — i.e., spend $1M to gain $1,200,000.)
It might go without saying, but I’ll do so anyway because UX has traditionally been an oppressed field where people have become accustomed to arguing that their work is so much more valuable than anything else that they deserve bigger budgets.
If or when we reach the state where UX has about the same ROI as any other business activity, that doesn’t imply that UX work shouldn’t be done. No more than it implies that those other company projects shouldn’t be done.
UX work should still be done, and significantly more of it than we currently do, but the expansion should stop when we reach the point where an extra dollar spent on UX is worth less than an extra dollar spent on some other business priority. This point is within reach, as UX has diminishing returns, a phenomenon common to most other business projects.
UX Is Normal, not Special
For many UX professionals, my overriding conclusion may come as a harsh adjustment to a new reality: UX and UX people used to be special, but now we have become normal.
25 years ago, this image could have symbolized a UX professional on the left and a user on the right. In 10 years, the same image might be used again with the labels swapped: happy users make for unhappy UXers who no longer feel special or will command their traditional high fees because of the stupendous ROI they once generated. (Leonardo)
For the first roughly twenty years of my career, I felt special because UX was oppressed — and even hated by people who resented the new restrictions on their creative design freedom. I still remember the Flash designer who emailed me that he wanted to punch me in the nose after I declared that Flash was 99% bad.
While it wasn’t pleasant to work in an under-resourced and underappreciated field, it did make me feel special. I was fighting for a good cause!
The last roughly twenty years of my career, until I retired from business in 2023, still made me feel special because usability endowed UX professionals with the special superpower of knowing what users wanted — something that no other disciplines in product development could do.
However, the very success of knowing how to increase online profits gradually changed usability from a narrow field of superheroes into a commodity business. When we know how to make money, of course, everybody wants to have that power. And the “secret sauce” of watching users isn’t that secret. I’ve always said that you can learn the basics in an hour, after which you just need to sit quietly with some customers and observe them using your product without biasing them with comments or instructions.
I won my 42-year battle for usability. Not to the extent that current designs are perfect but to the extent that most of what I fought for has become the norm. (If sometimes honored in the breach.)
UX is normal, not special. That’s good for users and for business, even if it doesn’t always feel as happy for us.
UX was special. Now it’s normal. Good.
I made a 3-minute explainer video summarizing this article: