Just one web site among those operated by the top 50 retailers in the Top 500 Guide—AbercrombieKids.com—loaded in 1.5 seconds or less throughout the entire shopping period from Thanksgiving through Cyber Monday, according to a new index from digital marketing analytics and privacy technology firm Evidon Inc. provided exclusively to Internet Retailer.
In contrast, six of those retail sites took on average three seconds or more to load each day in that period.
Three seconds is significant because if a site takes longer to load, 40% of consumers will abandon it, according to application delivery and security firm Radware Ltd.
Those slower sites, with their average page load time in the five-day period, are:
- LLBean.com, 3.2 seconds
- Macys.com, 3.9 seconds
- Netflix.com, 3.9 seconds
- Newegg.com, 3.1 seconds
- SaksFifthAvenue.com, 3.6 seconds
- ToysRUs.com, 3.7 seconds
None of the above retailers immediately responded to a request for comments, except Toys ‘R’ Us. “We take site performance very seriously and have taken a number of steps to ensure we continue to improve the customer experience in the short-term and further enhance our mobile capabilities over the long-term,” a spokeswoman for the retailer says.
One factor affecting load time is how many tracking tools there are on a web page to capture data about each site visit. That includes tracking software loaded by outside companies, such as advertising networks and analytics vendors. If they load slowly or with errors, they can cause latency, or a slowing, in the page load time.
More trackers doesn’t always equal more problems, though—notably for Amazon.com, which had between 388 and 461 tracking tools installed on each day of Thanksgiving weekend but hardly strayed from a 2.5-second loading time. That’s because Amazon doesn’t load all those trackers for every site visitor, but tailors its tracker-enabled data collection and ad serving for each shopper, says Andy Kahl, director of data analysis at Evidon. “Amazon is an aggressive re-targeter and is positioned to do some very sophisticated stuff based on behavioral data,” he says. “It doesn’t surprise me that they reach out for data from, and share data with, as many partners as possible to execute these strategies.”
Overall, however, the sites that logged the 10 slowest loading times in the five-day period had roughly 10 times more trackers installed than the sites logging the 10 fastest load times—105 trackers versus 11 trackers on average. For the 10 slowest sites and days, the median latency caused by trackers was 0.49 seconds, versus 0.26 seconds for the 10 fastest. See this chart for data on all the sites:
AbercrombieKids.com, which maintained speedy page loading throughout the Thanksgiving weekend, had an average of 10 tracking apps on its site each day—well below the average of 89 trackers for the 50 sites in the index. Included in those 10 was BrightTag’s tag manager, which is a single tracker used to replace multiple other trackers on a site and manage their data requests from another location, says Kahl says. Tag managers coordinate vendors’ and publishers’ data requests so they don’t all bombard a page at the same time and slow it down.
AbercrombieKids.com also uses a tracker from ad technology company OpenX that is “highly configurable,” meaning Abercrombie can manually finesse how it, too, handles publishers’ data requests in order to optimize site performance, Kahl says. “It appears that they really have their act together in terms of optimized architecture,” he says.
Just one web site among those of the Top 50 retailers logged both one of the 10 fastest and 10 slowest page load times over the Thanksgiving holiday weekend: Barnes & Noble Inc.’s BN.com, which loaded in about one second on Saturday but took nearly five seconds to load on Cyber Monday. However, that’s probably an anomaly, Kahl says. That’s because BN.com redirects automatically to BarnesandNoble.com, he says, which on Cyber Monday loaded in just under three seconds.
“My guess is that Barnes & Noble was doing a little bit of tracking to gauge how frequently the BN.com domain was the gateway to the flagship site,” he says, noting that BN.com had only three trackers on average over the weekend while BarnesandNoble.com had about 150. “But since it’s just a redirect, and we don’t have a ton of volume for it, my guess is that the high latency for BN.com is anomalous.” He did not say exactly how much volume, or web site traffic, Evidon measured for the two domains.
Evidon tracks the average page load times of the web sites of the Top 50 retailers in the Top 500 Guide, along with the number of third-party trackers deployed on each page and the average latency they cause in its load time, for Internet Retailer. The data come from roughly 10 million consumers who use Evidon’s Ghostery tool, a browser add-on that tells them which companies are tracking their online activity and for what purpose, and who have opted to participate in an anonymous panel to provide data about their site visits to Evidon.
The panel includes residents of more than 50 countries and users of every major browser—Mozilla Firefox, Google Chrome, Apple Opera and Safari, and Internet Explorer—with just under 5% of the data coming from two mobile platforms: Firefox users on the Android operating system and users of Evidon’s Ghostery browser on the Apple operating system. Evidon collects the data hourly. For the Internet Retailer index on a regular basis, Evidon will average the data for each web site over one week. For this holiday index, though, the data were averaged over each day.