The leading eCommerce websites have gone (or are in the process of going) headless. To keep up, eCommerce platforms have been working on ways to support modern, API-driven frontends. Some have invested in creating their own storefront frameworks, like Magento PWA Studio, SAP (Hybris) Spartacus, while others have invested in creating great API coverage, like Shopify. While every platform is different, speed should always be a priority. Especially in the days of COVID-19 and stay-at-home living.
Website speed impacts your entire digital strategy, from SEO ranking to traffic to conversions and revenue. As Google’s Lighthouse performance score has become akin with site speed, we analyzed the performance scores of the leading 500 US internet retailers (IR500) by eCommerce platform to assess which platform delivers the fastest sites. The surprising answer is: none. In fact, we found that in-house systems outperform the leading eCommerce platforms by 61%. Read on for our detailed findings.
How does the IR500 perform?
The average Lighthouse mobile performance score for the IR500 was 26.1 with a median score of 22 and a standard deviation of 18.9 points. This indicated a positively skewed distribution, where half of the leading 500 eCommerce sites by revenue scored 22 or less. On a percentile basis, 75% of the sites tested had a performance score of 37 or below, and 90% of the sites tested scored a 55 or lower.
The 90th percentile is interesting, as Google stated that a performance score of 50 marks the 75th percentile, when looking across their dataset. While the unique nature of eCommerce websites (e.g. dynamic pricing, real-time inventory, etc.) is known to slow a site down, the extent that the average eCommerce site is slower than other types of sites, such as publishers and other static sites, wasn’t known. Until now.
Among IR500 retailers, automotive websites saw the highest performance scores, outperforming the IR500 average by 40.57%. Sports & Hobbies websites scored an average that was slightly (0.23%) lower than that of the IR500. And Health & Beauty websites came in last, with an average performance score that was 25.27% lower than that of the IR500.
(Our methodology and more info about Google's PageSpeed Insights and Lighthouse Scores can be found at the end of this article.)
In-house systems outperform the leading platforms by 61%
The average Lighthouse performance scores for websites on the leading eCommerce platforms: Salesforce Commerce Cloud, Magento, Shopify, SAP Commerce Cloud (Hybris), and Oracle Commerce Cloud was 18.7 with a median of 15. This meant that the average performance score for sites on the leading eCommerce platforms was 28% lower than the IR500 average.
In contrast, the average performance score for websites using in-house / homegrown eCommerce systems was 15% higher than the IR500, with an average of 30.1. Perhaps this is one of the reasons that nearly a fifth of the IR500 websites use their own, in-house solution.
SAP Commerce Cloud (Hybris) seemed to be the least popular solution with only 9 domains. These sites received an average Lighthouse score of 36.5 with a median of 18 and standard deviation of 31 points. This wide deviation was due to two major outliers, Grainger and Purchasing Power, who scored 90 and 88, respectively. When excluding these domains, SAP Commerce Cloud’s average dropped to 21.6, with a median of 15 and standard deviation of 10 points.
Websites on Oracle Commerce Cloud saw an average performance score of 19.2 with a median score of 17. The tech giant, Apple, claimed the lead with a performance score of 60, while Best Buy scores a low 5. Magento and Salesforce Commerce Cloud (SFCC) had comparable scores, both averaging at roughly 18.5. However, SFCC held a higher median at 17, while Magento sat at 13.5. Lastly, Shopify websites underperformed the average for the IR500 sites by an astonishing 43.6% with a low average performance score of 14.7.
While the average website on these three platforms underperforms the average for the IR500 by almost 30% this doesn’t mean that sites on these platforms are inherently slow. For example, AKIRA on Magento and Planet Blue on Shopify scored a 74 and 67 in Lighthouse, respectively. Each of these scores are higher than the average for their eCommerce platform by 4-fold. Another example is Shoe Carnival on SFCC. The retailer scored a 70, which is 3.8x the average performance score for the platform. These three websites are examples of the performance scores you can attain with an instant website on Layer0.
Outperform the eCommerce giants
The average performance score for the leading US 500 internet retailers by revenue is 26.1. This score can easily triple with an instant site on Layer0. We’ve helped complex eCommerce sites with dozens of tags, real-time inventory, and dynamic pricing hit mobile performance scores of +80 and we can do the same for you.
At Layer0, our goal is simple: We want to make the Web instant. And we won’t stop until shoppers' needs are met, and our customers experience a substantial lift in their bottom-line. Websites on Layer0 see median page loads of 320ms (FCP), and organic traffic and conversion lifts in the 15-30% range.
Layer0 let’s developers squeeze every millisecond out of every speed optimization effort (client-side, server-side, and on the CDN level) through the combination of cutting edge tech, including portable frontends with built-in AMP and server-side rendering support, predictive prefetching, and a CDN-as-Code that streams and caches dynamic content at the edge so that your site remains 5 seconds ahead of shoppers’ taps.
In the wake of COVID-19, we understand time is of the essence. We can drop page loads down to 1-2 seconds in just one week, or down to 320ms medians (FCP) in a matter of weeks.
The data shows that faster websites are seen on in-house / homegrown solutions, rather than the leading eCommerce platforms: Salesforce Commerce Cloud, Magento, Shopify, SAP Commerce Cloud (Hybris), and Oracle Commerce Cloud. The average performance score for IR500 websites on these leading platforms was 18.7. In-house solutions saw scores over 60% higher than this.
While we found Lighthouse scores on these platforms to be surprisingly low, we also found that these sites aren’t inherently slow, and speed can be improved on the right infrastructure. Our Magento, Salesforce, and Shopify customers are seeing at least 4x the average Lighthouse score for their eCommerce platform, on Layer0. With a March 2020 update to Layer0, sites are able to achieve scores in the high 80s and 90s. Just take a moment browsing through TheTieBar.com, ShoeCarnival.com, AKIRA.com to experience instant for yourself!
We can get your page loads down to 1-2 seconds in 1 week, or can get your site sub-second in just a few. If you’re looking to increase organic traffic, conversions, and revenue, schedule a consultative conversation with a site speed expert to discuss your options.
In the first week of February 2020, we analyzed the mobile site speed of the homepages of all the primary domains of the leading 500 US internet retailers (IR500) by revenue, using Google’s PageSpeed Insights web performance tool. The IR500 serves as a reasonable snapshot of the best eCommerce companies in North America and includes known names such as Amazon, Apple, Williams-Sonoma, Overstock and more. Note that of the 500 companies in the IR500, 11 were excluded because their site was not directly transactional, the site was no longer live, or PageSpeed Insights was unable to analyze the page. At the time of our analysis, PageSpeed Insights was still using Version 5 of Lighthouse. A new scoring change is coming in Version 6, which is currently in Beta.
PSI analyzes a site using two underlying tools. The first is what’s called lab data (sometimes called “synthetic data”). This is generated by an automated process that loads the page in a browser site load on Google’s server and analyzes it with the open-source Lighthouse tool. Running Lighthouse through PSI essentially removes device-based variation in lab data. The second tool used is Google’s Chrome User Experience Report (CruX), which provides real-user measurements for site speed. The difference here is not only in RUM vs synthetic data, it is also in the devices used to test a site, which in the US tend to be faster than the PSI defaults (e.g LTE vs 4G)
For more about why Lighthouse scores as measured by PSI matter, check out this article.