Why did top UK retail sites end 2016 bigger and slower than ever? And what should they do about it?

Our final performance benchmark results of 2016 for the UK’s top retail sites revealed an all too familiar picture of bigger, slower web pages.

Load times and average page sizes have been creeping inexorably up since 2013. Over that time, we’ve noticed a few trends that might be contributing to the slowdown. Designs that favour big hero images, for example. The rise of third-party content. Or the simple tendency for the volume of content to be constantly ratcheted up – new styles and/or scripts being added more often than redundant material is cleared out.

Does it really matter?

It’s tempting to think that the answer is no. Most of us have fast broadband now. And even if we’re increasingly shopping on mobile phones, surely modern phones are more than capable of handling bloated web pages?

There are two problems with this.

One is that better devices, browsers and networks can only take us so far. Mike Belshe showed us back in 2010 that the law of diminishing returns kicks in at a fairly meagre 5Mbps when it comes to bandwidth – increasing bandwidth beyond this does relatively little to improve load times. Latency is arguably a more important bottleneck, and the laws of physics limit the extent to which we’ll ever be able to overcome that. What’s more, latency tends to be a more serious obstacle on mobile networks.

The other problem we’re facing is that despite (or in some ways because of) advances in technology, we are now much less able to predict the conditions in which people are accessing the Internet.

We design and build websites that work beautifully on high-end devices and fast networks. But what about people using cheaper smartphones? Or someone trying to access your website on a slow or unreliable network, on a train, for example? Even if your website is fast in optimal conditions, you could well find that there’s a long tail of visitors getting a very poor user experience. The graph below is from NCC Group’s Real User Monitoring (RUM) service. It shows average page load time per session for a UK retail site (broken down by device type). The shaded area represents sessions with an average page load time of over 3 seconds.

rum1aopt

There’s plenty of evidence now to link performance to KPIs such as conversion, bounce rate and, ultimately, revenue. So this is a long tail of missed opportunities, likely to include frustrated would-be customers who gave up after waiting too long.

The answer is – or should be – relatively simple. Assume that your customers are going to be on low-end devices with poor connectivity, and build your website with that in mind. Remember that memory and processing power are limiting factors just as much as bandwidth and latency. So keep JavaScript to a minimum, reserving it just for those tasks that can’t be done any other way. Be wary of large volumes of imagery. This doesn’t just take a long time to deliver over the network, it also has to be decoded and rendered – resource-intensive activities for low-end devices (especially if they also have to scale images to fit the viewport).

If all this sounds like a challenge, remember that we used to be able to do this. The lighter, faster websites that UK retailers were building in 2013 worked perfectly well. According to the HTTP Archive, average page size was well under 1MB back in 2010. It is now well over 2MB.

As much as this is a slightly depressing story from most UK retailers’ point of view, it also represents an opportunity – most might be slower than they should be, but some are bucking the trend. And they’re perfectly placed to snap up the frustrated customers who leave sites that aren’t optimised for all visitors.

Try RUM free to find out how many of your customers are seeing slow load times.

Leave a Reply

Your email address will not be published. Required fields are marked *