Lighthouse scores of the top 20 most visited websites

Motivation

Dealing with analyzing core web vitals of customers websites has become an habit at work. Yet, performance is a complex topic, both from a technical and organizational standpoint and it is sometimes hard to understand the business increment one can gain from a significant improvement in their page loading speed.

But beyond this, I wondered where the top players trafic wise were currently standing on that matter. I needed some content to kick start that site, so here we go!

Methodology

We will use similarweb ranking of the top 20 websites worlwide. We will perform the tests through PageSpeed Insights. One limitation here is that most of the pixels and third-party tracking calls would be blocked by Consent Management Platforms. But we aim to measure the performance for a new visitor so this should not matter. Another limitation is that pagespeed website may run the test from a server located in another country than the website usual trafic. As a result, latency may be significantly higher, biasing the results. Similarweb is not perfect but it still offers one of the most accurate websites trafic estimates out there. Tests simulate mobile experience, and we will focus on that device only for now. Finally, most of the websites at the top of the list are social networks or search engines, with limited capabilities displayed on the page (login or search form) and thus , limited need to load hundreds of resources. At least, that's my hypothesis, we will see how it turns out.

Results

You would find below the min - max - median scores for each core web vital metric, which we can interpret as half of the audited websites performed worse:

  • First Contentful Paint (FCP): 1.1 s - 5.3 s - 2.3 s
  • Speed Index: 1.2 s - 18 s - 4.2 s
  • Largest Contentful Paint (LCP): 1.6 s - 28.2 s - 4.9 s
  • Time to Interactive (TTI): 1.7 s - 28.8 s - 7.8 s
  • Total Blocking Time (TTB): 0 - 7540 ms - 350 ms
  • Cumulative Layout Shift (CLS): 0 - 0.408 - 0.032

If your site performs better, congrats! Time To Interactive seems crazy high, especially compared to the Total Blocking Time. Looking at the detail of the audits for websites with TTI over 7.5 seconds, if we stick to the Lighouse results recommendations, most of the issues are -without a surprise- associated with Javascript parsing and evaluation, monopolizing the main thread.