“40% will abandon a web page if it takes more than three seconds to load.”
This is a statistic I am seeing quoted more and more in articles over the last year or so. In fact, I’ve even used the statistic myself to back-up a blog post on mobile site performance. It’s a pretty compelling statistic and I can fully understand why it’s being quoted so frequently.
Essentially it’s saying that you have a very short amount of time to keep someone’s attention online. Consumer standards are high and a sluggish site with lengthy load times won’t cut it anymore.
I believe the sentiment behind the statistic is great. It tells sites that they have to adhere to a higher standard of performance, and that creating a bloated site full of heavy images and graphics, superfluous animations and flashy transitions goes against what users really want. It also sends a clear message to lazy site owners who have allowed their backend to become a mess of script and unused plugins that they need to clean up their act.
However the more I think about the statistic, the more it bothers me. Essentially it’s saying that if I sat down 10 people in a room and let them loose on a website, four of them would flat out abandon the site if a page happened to take more than 3 seconds to load, regardless of the type content they were trying to access, device limitations and fluctuating connection speeds. Who are these people that have no patience for web content that isn’t delivered to them in the same amount of time it takes to watch half a Vine?
The statistic in question is based on an Akamai study into the web page response times for e-commerce sites. So straight away the context has changed from any web page to any web page on an e-commerce site.
The study was conducted through an online survey and based the feedback of 1,048 shoppers in the US. Although I don’t know the complete specifics of how the survey was structured, there are a few issues with the way this study gathered its data.
• It’s based on opinion, not facts.
Asking a user how long they would expect a page to load is completely different to observing how a user interacts with a website. By asking a person what their expectations are, you are invariably going to be leading them into their answer. A better way to gain this insight would be to actually watch how users interact with sites to see where they grow impatient rather than simply asking them.
• The lack of context.
There is no real context to explain what the user was expecting to load other than saying that it was for an e-commerce site. Users expectations are likely to shift based on what type of content they are trying to access. For example, a user looking at a large product list might be more excepting of slower load times when compared to a webpage with typically less content like a checkout page.
• The sample size is too small.
If we are talking about the entire internet population, 1,048 online shoppers is an extremely small sample size. In fact, based on current worldwide internet usage, the sample size is around 0.000044% of the entire internet population! Obviously gathering a large sample size is is a common problem with conducting any kind of research, but in this instance it feels particularly small.
• The demographic distribution is biased.
While the age and gender distribution is fairly diverse; the survey only asked people based in the US. Internet speeds vary greatly across world, so what might be true for US users, may not be true for Korean users who average 60+Mbps broadband speeds, and certainly not true for users in South Africa who average around 2Mbps.
It’s worth noting that I think the implications of the study are actually good and should be taken on board by all sites. Decreasing your page load time is one of the best things a site can do to improve their bottom line. However it’s important the results of studies like this aren’t used as gospel and applied to all internet users, especially when the means of gathering the data are flawed and the sample size is so small.
It’s also important for marketers to look past statistics and focus on expectations of their own users. It’s easy for marketers to rehash stats to back-up their points, as opposed to actually gathering their own data or better yet, talking directly to their customers.