Skip to Main Content

The 2019 RSR Report on eCommerce website performance is now available. Download Today

Performance

Performance Benchmark: Automotive Sites

For this benchmark, we collected nearly a month of data on 10 major automotive websites.  The benchmark collects samples from four locations worldwide and averages them to provide the daily figures.

Overall Performance: Load Time

Below we see the daily averages of Time to Interact (a metric marking when the page content is visible and the user can interact with it).

Overall, performance on these sites is pretty dismal.  Look closely at the chart and you’ll see a yellow line just above 5 seconds on the y axis – that’s the average for all sites across the entire web.  No site on this benchmark had noticeably faster than average performance on any day sampled.

  • Motortrend.com stands above the rest (in a bad way) with Time to Interact averaging over 10 seconds on every day sampled. (Notice the blue line with inverted triangles). 
  • Other sites had worse performance on certain days, but it was isolated rather than consistent
  • It’s difficult to see with all the busy data, but Jalopnik.com had the best performance overall, with an average TTI of 6.33 seconds. 

Asset Count and Asset Weight

This matrix shows averages for all the data in the number of assets (a.k.a. requests — all the HTML, images, JavaScript, CSS and any other files on the page) the collective “”weight”” in bytes of those assets, and Time to Interact.

In a recent study of thousands of websites, we found that the content metric with the highest correlation to Time to Interact is the number of assets on the page.  Generally speaking, more assets means more round trips to the server to grab content, and more time for the visitor to wait for the page to be done loading. Most of the sites in this benchmark are in the bottom 20% of the web (meaning the highest number of assets) — not surprising given the sub-par performance we noticed.  The slowest site, Motortrend.com, also has the highest number of assets, with an average of 258 — a whopping number that places it in the bottom 2% of websites.  That’s a ton of content for a single page.

Is Time to First Byte a Factor?

A common thread among site owners is the idea that servers (their capacity and power) are a major determinant in performance.  This benchmark is a great case study showing why that’s largely an outdated concept.  Notice the web-wide average among this segment for Time to First Byte, a metric that’s driven mainly by how fast a server can respond with the first byte of a website (middle mile latency is also a factor).

The sites are largely BETTER than average for Time to First Byte, and the ones that are worse than average still fall well into the realm of acceptable performance. (The red shading indicating bottom 25%, which you can see in the first chart on this post, is not even in the picture in this chart).  These are all popular sites with, one can assume, money to spend on improving their sites.  They clearly are not slacking on server performance — and yet, when it comes to the actual user experiences, they’re among the slowest on the web.

This underscores the point that on the modern web, 80-90% of a page load is consumed by work done in the browser, not by server latency or delivery over the middle mile. “”Upgrade servers”” is rarely the best answer to performance problems.

Summary

Looking at the automotive sites, we found:

  • Without exception, the sites had worse than average performance – Motortrend.com had the longest average load times; Jalopnik.com had the fastest performance
  • Many of the sites were in high percentiles for number of assets and page weight, which likely contributed to performance
  • The pages scored suprisingly well for “”Time To First Byte””, ruling out server performance as a factor in overall performance

Want Us to Cover Your Industry?

Leave a comment or send an email to info@yottaa.com to let us know what sites you’d like to see in a benchmark.

Don’t let slow site performance cost you conversions.Let's Talk