User experience and loading times
First, let’s start with the question the study set out to answer: how does the predictability of loading times affect the user experience and the task at hand? Here’s the experimental task designed for the occasion: in a purpose-built mail client, messages came in. Subjects were asked to clear out the spam, then forward to their manager, as fast as they could, the emails meant for the manager. A fairly simple task, then.
After each interaction, and as with any standard computer system (especially on Windows, some will say, but that’s not today’s debate), a loading time was needed to display the next window. There were two conditions:
1. highly variable loading times, but generally short,
2. barely variable loading times, but generally long
And what about time?
The subjects’ interaction time (loading times excluded) was measured. Then, after each phase of the experiment, they rated their workload with a questionnaire that’s standard in the field (the NASA Task Load Index). A second questionnaire (AttrakDiff) captured their user experience.
The experiment came to the following results:
- the subjects’ performance (information processing time) is significantly better when the loading times are barely variable, even if this means an increase in the overall waiting time; (condition 2.)
- despite this, the user experience and the workload are not affected. Neither by the variation in the overall waiting time nor by the variation in the variability of that waiting time.
So then, what does all this tell us? Well, it takes us straight back to the criteria for the ergonomic evaluation of interfaces. The classic ones (Bastien & Scapin’s criteria in France, Nielsen’s heuristics in the US) all talk about the need to keep an interface consistent (homogeneous): the “Cancel” button doesn’t jump from the left to the right, the buttons aren’t flat design in one place and skeuomorphic in another, clickable links aren’t highlighted in blue here and underlined in red there…
For all of us PC users who’ve been put through the wringer by sites and interfaces that just won’t behave, this is plain common sense. But the fact that we’re so often let down by the ergonomics of the systems we’re trying to use shows it’s not nearly as obvious at design time as it sounds. Before implementing an interface, it’s often useful to test it with prototypes that make it possible to simulate the interactions and check the consistency of the user journeys.
Is loading time a fundamental factor?
This result is one more reminder that the user experience isn’t just the look of the interface: it’s the whole experience of interacting with it. Loading time is part of that, so consistent loading times are one of the things you can give users (or operators) to help them perform better without spending any extra mental effort. An improvement on the cost/performance/risk triad.
To be thorough, we should also come back to this figure of 10 seconds (or 5 seconds, or 13 seconds…). Past that point, you risk losing the user for good. Most of the studies behind these figures are a little dated (Nielsen, 1994, the dawn of the internet). They often deal with the wider public, particularly in e-commerce, with no animation to keep people occupied while they wait.
In short, loading time is a fundamental factor in the user experience. And far be it from us to suggest it shouldn’t be optimised. But not all interfaces are websites, and not all websites are online sales platforms. Operators will be far more forgiving and patient with a business interface running complex calculations and lightening their work than with an e-commerce site that has plenty of competitors. To learn more about optimising your interfaces, feel free to read our page on user tests, which make it possible to measure the real usability of your products.
As is so often the case in ergonomics, then, the truths aren’t as universal as we’d like to think, and that’s exactly where an ergonomist’s expertise earns its keep! If you want to assess how your interface performs, an ergonomic audit will help pinpoint usability problems and suggest the right recommendations.
In summary
when time is a performance factor for your operators, the consistency of loading times (even by levelling them up) can make them more efficient,
in the other cases, some useful figures to know today are as follows:
- below 0.1s, it feels instantaneous,
- below 1s, the user doesn’t really perceive a transition,
- beyond 10 seconds, it’s hard to keep the user’s attention on their original task.
To go further on these questions of performance and optimisation, discover our approach to ecological design and eco-design, which aims to create lighter and faster interfaces.
Reference
Weber, F., Haering, C., & Thomaschke, R. (2013). Improving the Human–Computer Dialogue With Increased Temporal Predictability. Human Factors: The Journal of the Human Factors and Ergonomics Society, 55(5), 881-892.
Nielsen, J. (1994). Usability engineering. Elsevier.
