During my fourteen years in the web performance industry, I have delved into, written about, and presented a wealth of content on the psychology of page load speed—in other words, why we are so eager to have a fast and smooth online browsing experience. In fact, the first chapter of my book, Time is Money (reprinted with the kind permission of O'Reilly Media), is entirely about this topic.
Recently, I shared some research findings that I am very fond of at the Beyond Tellerrand conference (related video link), and I wanted to summarize them in an article. We will discuss:
- Why time is a key and often overlooked usability factor
- How we perceive wait times
- Why our memories are not so reliable
- How the end of an experience can have a huge impact on our perception
- How fast we expect pages to load (and the reasons behind it)
- The meaning of "flow" state in web browsing
- How delays affect our productivity
- What we can learn by measuring "web stress"
- The impact of slow loading speeds on our overall perception of brands
We have a lot of exciting content to share, so let's get started!
Time is a Key Usability Factor#
If you do not consider time as a key factor in user experience, then you are overlooking a fundamental component of user experience.
I must admit that during my previous career as a user experience tester, I tested countless websites in a lab environment without ever considering the importance of page rendering time.
To be fair, that was in the early 2000s, when website loading speed was hardly a concern. Fortunately, things have changed dramatically since then. Now we have a wealth of research explaining why waiting is so difficult—this explains why website loading speed is so important—not just from a business perspective but also from the deeply rooted neurological perspective of our brains.
Let’s take a broader view of how humans handle various forms of waiting.
Time is an Indispensable Element of User Experience#
If you do not consider time as a key factor in user experience, then you are overlooking one of its core components.
Looking back, during my previous career as a user experience tester, I tested countless websites in a lab environment, and embarrassingly, I never took page rendering time into account.
Admittedly, that was in the early 21st century, when webpage loading speed was hardly a focus. Fortunately, since then, the situation has changed dramatically. Nowadays, we have a wealth of research explaining why waiting is so difficult—this also explains why website loading speed is important—not just from a commercial perspective but also from a deeper neurological perspective of our brains.
Now, let’s broadly explore how humans cope with various forms of waiting.
How Do We Perceive Wait Times?#
In short: we do not handle it well.
Queueing theory is the mathematical study of waiting lines (whether actual or virtual). Digging deeper into this field, you will find some extremely interesting stories. One of my favorites occurred at Houston Airport.
The airport's customer relations department received numerous complaints about passengers waiting too long for their luggage at the baggage carousel. The airport management attempted to resolve this issue by increasing the number of baggage handlers, which reduced the average wait time to seven minutes. However, the number of complaints did not decrease.
What was the solution? They did not add more baggage handlers; instead, they made the distance between each flight's arrival gate and the assigned baggage carousel as far apart as possible. As a result, although the time it took for passengers to walk to the baggage area increased sixfold, the average wait time at the carousel decreased to one minute.
Subsequently, passenger complaints nearly dropped to zero.
The takeaway from this experiment is:
- Waiting is an unbearable experience
- Passive waiting feels even more difficult
- The perception of speed is more important than actual speed
These principles apply to nearly all waiting scenarios, including waiting for web pages to load.
Our Memories Are Not Always Reliable#
Our perception of time can be influenced by various factors, including our age, location, emotions, and various external stimuli. Unsurprisingly, this inconsistency in perception also applies to our online experiences:
On average, internet users feel that web page loading times are 15% slower than they actually are. When recalling this experience, they believe the loading time was 35% slower than it actually was.
People generally believe they spend 9 minutes a day waiting for slow-loading web pages, which adds up to wasting two days a year. (Even if this data may not be precise, it interestingly reflects people's feelings about their online experiences.)
Adding cues like loading spinners and progress bars can make us mistakenly believe that page loading speeds are 10% faster than they actually are. We not only feel that waiting is slow, but the experience also seems slower when we recall it afterward.
The Final Stages of an Experience Have a Huge Impact on Perception#
The "colonoscopy effect" comes from a study where two patients recorded their pain levels during a colonoscopy.
From the data chart below, you can see that although Patient A had a shorter examination time and similar peak pain levels, they perceived their experience as longer and more painful. This conclusion illustrates that because Patient A's experience ended with a peak of pain, this pain profoundly affected their perception of the entire experience.
What does this finding mean for website performance? If your website loads quickly for most of the time users are browsing, but becomes slow and unstable during the final critical stages—like during checkout—users may develop a negative impression of your website's overall speed.
Is this feeling fair? Probably not, but that is how our brains process information.
How Fast Do We Expect Web Pages to Load?#
While we claim that our expectations for online experiences are neither precise nor consistent, our actual reactions to different page loading speeds are quite consistent—this has remained unchanged over decades.
In 1968, Robert Miller published a study titled "Response Times in Human-Computer Dialogues." In the study, Miller noted that waiting longer than 2 seconds can disrupt our attention and affect productivity.
Research conducted by usability expert Jakob Nielsen in 1993 and 2010 found that:
- A response time of 0.1 seconds creates the illusion of instantaneous feedback
- A loading time of 1 second allows our thoughts to continue smoothly
- A wait time of 10 seconds is barely enough to maintain our attention
- Beyond 10 seconds, our attention begins to drift, and once the page loads, it becomes more difficult to refocus.
The internet is changing, web pages are evolving, but user expectations remain constant. Data on human perception and response times has remained consistent over the past 45 years. This data reflects our inherent reactions, which we cannot control, and they are the same regardless of the type of device, application, or connection we use.
But why is this the case? This leads to some very interesting points.
Nielsen pointed out that people's reactions to slow loading are attributed to two aspects of our brain function:
- Our poor short-term memory—information in short-term memory fades quickly.
- Our need for control—being forced to wait makes us feel powerless and frustrated.
Our Impatience is Innate#
Our impatience is an inherent part of our remarkable human brain circuitry. At any given moment, your brain is processing three types of memory:
- Sensory memory
- Short-term memory
- Working memory
(There is also long-term memory, but it is not critical here.)
Sensory Memory#
Every time you see something, the visual information is captured by the photoreceptor cells in your eyes and sent to the occipital lobe of your brain. This constitutes your image memory. This is just one of your three types of sensory memory. (The other two involve auditory and tactile senses.)
Research on how image memory works has been ongoing for nearly 300 years. In an early study, a glowing coal ball was fixed to a wheel, and as the wheel spun faster, observers eventually saw a continuous light circle. The study concluded that the glowing coal ball must complete a full cycle in 100 milliseconds or less to create the illusion of a continuous fire circle.
Early research revealed what we now call "visual persistence," based on the fact that our image memory can retain visual information for about 100 milliseconds. Beyond this time, the memory "warehouse" depletes, and image memory needs to be refreshed with new visual information. This data has remained unchanged for centuries.
Interestingly, perhaps not coincidentally, 100 milliseconds is the page loading time target set by Google.
Image memory, along with the other two types of sensory memory, belongs to primitive memory. We cannot consciously choose which information to store, nor can we will it to last longer. (If we could, we might mentally collapse from information overload or inadvertently walk into danger.)
Of course, some sensory memories will persist... provided they can be quickly utilized and eventually transferred to our long-term memory.
Short-Term Memory and Working Memory#
If our sensory memory is responsible for providing comprehensive information about all our sensory experiences, then the task of our short-term memory is to filter relevant information and send it to our working memory for processing. Your short-term memory can hold information for a maximum of 10-15 seconds—just enough time for working memory to process, adjust, and control that information.
Thus, the goal of reducing perceived page loading time to 100 milliseconds is to:
- Prevent information from being lost from our image memory
- While also giving our short-term and working memory enough time to process the necessary information until it begins to fade.
What is "Flow," and What Does It Mean for How We Use the Web?#
Over hundreds of thousands of years of evolution, humans have developed the ability to perform actions in a graceful, orderly flow. Our daily tasks—such as making fire, hunting, toasting bread, and milking cows—are composed of a series of small actions that seamlessly connect to the next step.
In the book "Finding Flow: The Psychology of Engagement in Everyday Life," renowned psychologist Mihaly Csikszentmihalyi observed that those who regularly engage in seamless, sequential activities are happier than those who do not. He used the term "flow" to describe this state.
Over the past 40 years, as computers have permeated our homes, workplaces, and even our pockets, we have placed new demands on our brains. As many of us have experienced, unlike smooth sequential actions, computer use often comes with delays, downtime, and restarts.
In short, our brains, which seek fluid experiences, are not well-equipped to handle these interruptions in human-computer interaction.
Skepticism about the impact of delays, downtime, and restarts on productivity and other performance metrics has led some to believe that most people can actually adapt to poor performance. While these views may be somewhat correct, they may overlook the core of the issue.
Do Delays Really Undermine Productivity?#
A 1999 study explored the impact of interruptions in the workplace on workers, who experienced various disruptions during their daily tasks. The study examined:
- Their productivity,
- And their self-reported mental state.
Although the study primarily focused on general workplace interruptions and only partially addressed human-computer interaction, some of its findings are highly relevant to web performance:
1. Finding One: Participants developed effective strategies to cope with interruptions and maintain productivity.
The research showed that, at least for some workers in certain work environments, they not only learned how to cope with interruptions but may even have tried to compensate for potential performance declines.
2. Finding Two: However, this coping mechanism came at a higher psychological cost.
Overall, interruptions had a negative impact on mood and well-being. Additionally, participants ultimately needed to exert more effort to complete the same tasks.
3. Finding Three: Over time, interruptions affected participants' ability and willingness to resume work and take on new tasks.
Interruptions seemed to have a cumulative effect. As the number of interruptions increased, the time required to resume work became disproportionately longer. Participants appeared to lose motivation, and mental fatigue accumulated.
What Do These Findings Mean for Web Performance?#
When dealing with application delays, people may develop coping strategies to maintain productivity in the short term. But what is lacking is flow. Without flow, our motivation and sense of well-being will ultimately suffer.
Importantly, we need to remember that application performance is just one part of a broader world. Our daily lives are filled with events—from traffic jams to supermarket queues—that challenge our need for smooth experiences.
Slow websites are just one of many issues, but for those of us who work and live online most of the time, they add extra friction to an already friction-filled world. These effects are cumulative, as most of us cannot isolate stress.
"Web Stress" is Measurable#
When websites perform poorly, our reactions are very negative. (Some studies even suggest that browsing slow websites increases our blood pressure!) Given our deep desire for smooth experiences, this is not surprising.
In 2011, CA Technologies commissioned customer experience consultancy Foviance to conduct a series of laboratory experiments at Glasgow Caledonian University. Participants wore EEG caps to monitor their brainwave activity while performing everyday online transaction tasks. They either used a 5 MB network connection or one artificially throttled to 2 MB.
The brainwave analysis from the experiment revealed that participants needed to concentrate up to 50% more attention when using slower connections. When asked about their likes and dislikes regarding the websites they used during the study, participants frequently mentioned speed as their primary concern:
"The website was very slow, and it took a long time to load book previews."
"What I disliked most about this website was its speed."
The research also found that during the following stages of the transaction process, people were most likely to experience the highest levels of stress:
- Searching
- Finding and selecting products
- Checking out
- Entering personal information and completing the transaction
This is easy to understand. Online shopping is inherently stressful, as most of us strive to find the right products at the best prices. The checkout process—when we enter personal and credit card information—also carries a certain level of stress. In the case of slow-loading web pages, it is easy to see why the online shopping experience can become unpleasant.
Mobile Users Also Experience "Web Stress"#
Based on the desktop neuroscience study conducted by CA Technologies, Radware conducted a similar study on mobile device users in 2013.
(Disclosure: I was working at Radware at the time and guided this study. To ensure impartiality, the research and analysis work was outsourced to a third-party neuroscience research company called Neurostrata.)
The mobile stress study employed an innovative combination of eye-tracking and EEG technology to monitor the neural activity of a group of mobile users who were asked to complete a series of online transaction tasks using their devices. (The image below shows one of the study participants.)
In this study, participants were asked to complete standardized shopping tasks on four e-commerce websites using their smartphones. Some participants accessed the pages at normal speeds via WiFi, while others experienced consistent speed reductions (with software creating a 500-millisecond artificial network delay).
Participants were unaware that speed was one of the variables being tested; they believed they were participating in a study on general usability/brand perception.
Some key findings from the study included:
- Users experienced up to a 26% peak in frustration at critical moments.
- Consistent with CA Technologies' research, the browsing and checkout stages were the most common points of peak frustration.
- Faster page loading speeds were associated with increased user engagement. (This is a positive outcome!)
- Even under ideal mobile browsing conditions, all users experienced some degree of "web stress."
How Slow Loading Affects Our Overall Perception of Brands#
Yes, this also includes non-performance aspects of websites, such as content, design, and navigation.
After conducting the aforementioned mobile stress study, we conducted exit interviews with participants to ask for their impressions of the websites and companies. We input the adjectives from all interviews into a word cloud generator and created word clouds for each website's normal and slow versions.
Reminder: The only difference was the speed of the website. Since this was a blind study, participants were unaware of the speed differences. The results indicated that slower page speeds negatively impacted brand perception overall.
We found that slow pages damage the overall brand image.
Here is the word cloud generated after participants used the website at normal speed:
And here is the word cloud generated after experiencing 500 milliseconds of network delay on the same website:
While both word clouds contain positive and negative descriptive words, importantly, the word cloud for the slow website contains nearly three times as many negative adjectives as the fast website. The adjectives shifted from primarily "easy to use" (in the first word cloud) to a series of negative associations (in the second word cloud)—simply due to the delay in page loading.
Although some participants did notice a slight decline in performance ("slow" and "lagging"), they also formed negative impressions of other aspects unrelated to speed. They reported that the website seemed "boring," "unattractive," "clunky," "tacky," and "difficult to navigate."
In other words, slow page loading affected people's perceptions of three important aspects of the website that were completely unrelated to loading time:
- Content ("boring")
- Visual design ("tacky" and "confusing")
- Navigation usability ("frustrating" and "difficult to navigate")
Conclusion#
There is a striking gap between what we claim to want and what we truly need at a deeper level in our online experiences.
User surveys over the past decade show that our stated needs have changed over time—from an 8-second loading time in 1999 to 4 seconds in 2006, and now to about 2 seconds today. If we believe these surveys, we might think of ourselves as an increasingly impatient and intolerant species. We may tend to judge (or pity) ourselves as victims of the fast-paced modern life.
However, neuroscience research—exploring how we actually receive and respond to visual information—tells a completely different story. For decades, research has repeatedly confirmed the same result: overall, when our websites and applications (and technology in general) can respond within fractions of a second, our satisfaction and well-being reach their peak. We may learn to adapt to slower response times, but this adaptation is always, or at least for the foreseeable future, awkward and uncomfortable.
Yes, making your website faster does have business reasons. But caring about web performance goes far beyond commercial considerations.
As technologists—and empathetic humans—our goal should not merely be to provide a good enough online experience. We should strive to deliver a frictionless and enjoyable online experience, leaving visitors happier when they leave our websites and applications than when they arrived.