The newly released annual look at the digital world from online and mobile measurement firm comScore makes it quite clear that retailers better be ...
When traffic is heaviest, m-commerce site page load times can slow by 20% from non-peak traffic periods, but no more, says mobile and web performance firm Keynote. TV retailer HSN maintains very consistent performance throughout the day, a key to it jumping up the weekly Keynote Mobile Commerce Performance Index.
It certainly is difficult for retailers to offer the exact same mobile commerce web site performance 24/7. If servers are being pinged by a few hundred shoppers at 4 a.m. and tens of thousands of shoppers at 4 p.m., the added strain in the late afternoon is bound to have some effect on how quickly that retailer can load its m-commerce site on shoppers’ smartphones. What happens is the number of unusually long load times, or outliers, increases, bringing down the average load time, which is much shorter than the outliers. During non-peak traffic periods, with less server strain, there typically are far fewer outliers.
TV, web and mobile retailer HSN Inc. last week managed to maintain extraordinarily consistent m-commerce site home page load times that did not fluctuate—meaning very few outliers—far from the average of 4.69 seconds, reports mobile and web performance monitoring, testing and analytics firm Keynote.
“Every m-commerce site needs to contend with increased site traffic during peak hours of the day, and Keynote recommends that retailers aim for peak period slowdowns of 20% or less as their servers handle the increase in traffic,” says Keynote mobile evangelist Ken Harker, who adds that HSN’s peak period slowdown was far less than 20%. “Improving this metric can have big dividends in improving overall average page load times, and HSN is a great example.”
Typically, Keynote defines off-peak hours as midnight to 8 a.m. Eastern for North American sites and peak hours as 8 a.m. Eastern to midnight. Sites that exhibit good load handling have fewer slow data points during peak hours, Harker explains.
“Our general rule of thumb is that a site should not slow down more than 20%,” he says. “So, if the home page is taking 2.5 seconds to load at 4 a.m., you would not want to see the performance slow down to slower than 3.0 seconds at 4 p.m. If a site is regularly slowing down more than 20% every day, that’s a sign that the site has a capacity issue. Retailers should worry if that’s the case because a site that is struggling in a normal week is vulnerable to major load handling issues if a marketing campaign succeeds and suddenly you get twice the normal traffic, for example.”
For the week ending March 16, HSN jumped up five positions on the Keynote Mobile Commerce Performance Index to No. 6. Its m-commerce site home page loaded on average in 4.69 seconds and did so completely and successfully (known as success rate) 98.96% of the time. Weighting and then combining load time and success rate earns HSN an index score of 897 out of 1,000. The mobile home page contains 23 page elements (such as images and scripts) that combined weigh 158 kilobytes.
"HSN takes advantage of page caching and dynamic site acceleration via our third-party content delivery network to ensure consistent performance at any time of day," says Ed Deutscher, vice president of digital applications. "This ultimately allows us to provide an optimized experience to our customers regardless of their geographic location and when they choose to engage."
B2B powerhouse W.W. Grainger Inc. topped the index with a load time of a swift 2.75 seconds and a success rate of a perfect 100% for a score of 993. Its mobile home page contains only seven page elements that weigh 82 kilobytes. Keynote advises retailers wanting optimal performance to limit mobile page elements to 10 or 12 and limit page weight to 100 kilobytes or less. Sears Holdings Corp. came in second with a load time of 2.48 seconds and a success rate of 99.64% for a score of 987. Its mobile home page contains seven page elements that weigh 63 kilobytes.
The average load time for all 30 retailers on the index was 7.73 seconds, the average success rate was 98.81% and the average score was 799. The average mobile home page contained 40 page elements and weighed 420 kilobytes.
Two of the 30 retailers on the index offer responsive design sites, which use a single code base, a single set of web content and responsive techniques to render versions of the site that fit the size of the screen on a device a consumer is using. This saves a retailer having to build separate sites for desktops, tablets, smartphones and other devices, and delivers content from a single URL, which tends to strengthen search engine rankings. But the chief complaint about responsive design sites is slow load times, and that is the case with the two responsive sites on the Keynote index.
Fab.com came in at No. 26 on the index with a load time of a whopping 16.71 seconds and a success rate of 99.45% for a score of 584. Its home page for smartphones contains 73 page elements and weighs an unwieldy 2.1 megabytes. Fathead.com came in at No. 30 with a load time of 18.15 seconds and a success rate of 85.98 seconds for a score of 0. Its home page for smartphones contains 105 page elements and weighs 2.2 megabytes.
Click here then click on Keynote Mobile Commerce Performance Index Part 1 and Part 2 to view complete results for all 30 retailers on the index. Keynote measures, exclusively for Internet Retailer, 28 standalone m-commerce sites optimized for smartphones and two responsive design sites, which are single sites that render content in ways that fit the screen size of a device, including desktop PCs, tablets, smartphones and smart TVs. For the index, Keynote measures the smartphone versions of the responsive sites.
The 30 representative sites include merchants in multiple categories and channels, and of multiple sizes, ranging from such giants as Amazon.com Inc. to mid-sized retailers like Toolfetch.com. Keynote tests the sites in the index every hour Monday through Sunday from 8:00 a.m. through midnight EDT, emulating two different smartphones on two different wireless networks: Apple Inc.’s iPhone 5 on AT&T and the Samsung Galaxy SIII on T-Mobile, both using 4G networks. Keynote runs the tests in New York and San Francisco.