Download Speeds And Usability

Simply put, download speeds matters – it always has mattered, and it always will matter. Just visit an Internet café and watch web users in action. Before long you’ll notice how often the back button is used by those too impatient to wait for content, particularly when much of it will be available elsewhere.

In a recent interview with usability 'guru' Jared Spool, amongst the usual excellent thoughts on usability and the state of the industry, I came across the following remarkable statement:

"when we actually compared the usability of sites to their download times, we didn't see any correlations. None, zero, zip."

In other words, download times don't matter - it has no effect on the usability of a site if a page takes one second, one minute or one hour to download. Coming from a member of the usability community with a significant profile, someone who is a de facto representative of the industry, that's a pretty absurd claim.

Obviously it isn't true (aside from anything else, have we given up on 'efficiency' as one measure of usability?). Presumably the research methods used to arrive at such conclusions are flawed in some way - or alternatively Jared is so keen to convey the importance of other factors than simple speed of download (a noble aim in itself) that he is willing to inaccurately dismiss download speeds as completely irrelevant. Either way, this kind of statement is hardly a good advertisement for the usability industry.

Why Download Speed Matters

Given the debate concerning download speeds is obviously not going away, it might be worth revisiting the well-rehearsed arguments surrounding download speeds, if only to re-affirm their central importance to both usability as an abstract 'measurement' of site performance, and also business success online.

Simply put, download speeds matters - it always has mattered, and it always will matter. To argue the extreme position first, it is clear that if a homepage takes, for example, 5 minutes to load, there will be a significant drop off in the number of users willing to wait in order to see the site. I don't think it is possible to argue with that, and all research to date has consistently found that slower downloads mean more users giving up on a site.

You don't have to take anyone's word for this of course. Just visit an Internet café and watch web users in action. Before long you'll notice how often the back button is used by those too impatient to wait for content, particularly when much of it will be available elsewhere.

It is this 'competitive' nature of the web that can often be missed in a test procedure. If you are instructing a participant to buy a particular item on a particular site, of course they will persevere with the task in hand. Unless a critical error arises, 'fall off' will not be observed and can only be inferred from user comments. In the real world, where users are free agents, fall off due to slow download times is a serious problem.

Ensuring Fast Downloads

The only real rule when discussing download speeds is that 'faster is better'. There are so many variables, in terms of the type of content that is being presented or the differing ways in which users can perceive the passage of time, that to insist on arbitrary time limits is probably counter-productive.

Download speeds need to be 'fast enough', and fast enough is something that can only be measured in real world conditions, or in experiments that reflect real world conditions as much as possible. As a first step, test sites across multiple modem speeds and at several times of day. What may be lightning quick in your own office may look very different on the end of a 56k modem.

When user-testing sites for download speeds, try and allow users to give up or look elsewhere for information. Make the test as open as possible rather than forcing users to follow one specific route to information. Above all, don't push users through a test and then ask them to rate download speeds after the event. This is likely to produce results that are influenced by the quality of the rest of the site - something a user backing out due to slow downloads would not normally have seen.