Over the past few years, ecommerce sites have made a concerted effort to improve their shopping experience. Most ecommerce sites employ third party gizmos to add features like chat, recommendations, reviews, personalization and localization. The problem is, these third party scripts extract a cost: inconsistent website performance.
The Growth of Third Party Services
Figure 1: Growth of Third Party Service Usage
What Are Third Party Scripts?
How Third Party Scripts Slow Performance
Third Party Performance Metrics
While website outages during high traffic periods garner the headlines website slowdowns are ten times more likely than outages (Godskind 2010).
Gallino et al. (2018) found that a 10 percent decrease in website performance leads to a 2.6 percent decrease in decrease in retailers' revenue and a decrease of 0.05 percentage points in conversion. Gallino found a 10 percent increase in the number of hosts increased load times by 1.78 percent. Gallino and company also found that decreased website performance impacts revenue and conversion rates more for mobile than desktop.
The average webpage requires 90 to 98 HTTP requests to load as of Jan. 1, 2019 (HTTParchive.org). The average ecommerce site uses 85 third party services, so one web page can require up to 300 HTTP requests to over 100 servers to fully load. By the time the page loads the poor shopper can lose patience and leave the site.
Third Party Services Increase Performance Variability
While research has found that the number of objects on a web page is strongly correlated with page load times, the variability of load times is strongly correlated with the number of servers required to load a web page. Third party services by definition are deployed from external servers so they not only increase load times but increase the variability of load times. Variable load times are harder to "attune" to according to research from Chris Roast "attunability" studies (Roast 1998).
One reason why third party services increase response time variability is that external servers can have inconsistent response times themselves. By replacing third party services with first party services on the origin server, companies can improve load times and more importantly decrease and better control performance variability.
Analyzing Third Party Services
There are tools available to help quantify and analyzie third party services in websites. Speedcurve.com uses Web Page Test over time to create "speed curves" of websites. Speedcurve also created a proprietary third party analysis module specifically to analyze third parties. Webpagetest.org can show and exclude third party services with some additional work.
The use of third party services is growing at a rate of nearly 42% per year. The reliance of third party services is increasing load times (number of objects) and performance variability across the Web (number of servers). This trend is especially important for eCommerce sites where a 10 percent decrease in website performance leads to a 2.6 percent decrease in retailers' revenue and a decrease of 0.05 percentage points in conversion. To improve performance minimize the use of third party services, roll your own on-domain equivalents, or employ lower impact third party services instead.
"A 10 percent increase in load time leads to a 0.03 percentage point decrease in conversion in the mobile channel, while it decreases conversion in the desktop channel by 0.05 percentage points. Average conversion levels are 1.7 percent and 3.8 percent for the mobile and desktop channels, respectively. Thus, although the speed increase causes a smaller percentage point decrease in the mobile channel, the relative decrease in conversion in the mobile channel is larger relative to the desktop channel."
Found a 10 percent increase in the number of hosts increases load times by 1.78 percent. They recommend that companies should carefully select their third party content provides and reduce the number of hosts to optimize website performance.
By website optimization on 1 Feb 2019 PM