Don't use Page Load Time to predict conversion

Don't use Page Load Time to predict conversion

I've seen pagespeed audits where Google Analytics' Average Page Load Time was used to determine bad performing pages and claiming the webshop was missing revenue. However, pagespeed is more nuanced.

Google's PageSpeed Insights already is around for a very long time. But before Google introduced Lighthouse and it's metrics, PageSpeed Insights would show you First Contentful Paint (FCP) and DOM Content Loaded metrics as experienced by real users. Basically because of the lack of better metrics at the time.

A better perspective on real user experience

I still consider FCP a good metric towards first or early user engagement. But a lot of new metrics are introduced since then. Not to annoy developers, SEO specialists and other stakeholders, but to get a better understanding of the impact of code produced by us (both developers and platform builders) on real users.

And to be honest, Google did a pretty good job with Lighthouse and later on with introducing Core Web Vitals.When meeting the Core Web Vitals thresholds, users are 24% less likely to abandon page loads.

Sure, Lighthouse metrics keeps changing and characteristics of metrics like LCP and CLS already changed as well. But only to get a fair picture and better understanding of real user experience.

Page Load Time and conversion

But if these metrics are playing a role towards bounce and eventually conversion, why shouldn't we be using Page Load Time? The simple explanation: users don't care about when the page is fully done loading, or when the browser parser reached the end of the HTML document. Because:

  • it's quite likely that first noticeable changes (FCP) already happened way before Page Load Time;
  • and the hero image (LCP) was already displayed before Page Load Time.

Perceived performance matters

These are two examples of moment in time that are perceivable by users. That's why next to pagespeed or Core Web Vitals, you'll often be reading about "perceived performance" as well.

Obviously, most visitors won't familiar with the pagespeed terminology. But Page Load Time actually isn't telling anything about visual loading process at all. As a result, it isn't likely there will be a correlation between Page Load Time and bounce or conversion (which could be revenue of sign-ups, for example).

It isn't just perceived performance that matters

Although talking about perceived performance and visual loading process, there are some other metrics too that even became part of Core Web Vitals. And they could impact bounce and conversion as well. For the following reasons:

  • There shouldn't be too many layout shifts that could annoy or disorient visitors. This is part of the CLS metric.
  • And when trying to interact with the webpage by adding a product to the cart or opening the menu, the page should be able to respond right away to prevent delay's and eventually user frustration. This is part of the FID metric.

These metrics actually are likely to have a correlation with bounce and conversion. Contentking and listed some pagespeed and Core Web Vitals related case studies.

Average Page Load Time could still be of value

Depending on the way websites are build, you could miss out on revenue when having a page load below 1 second. Or have perfect user experience and bounce rate while having a Page Load Time of 9 seconds. The real user experience depends on the structure of the HTML and the priority of critical resources.


Don't hesitate to improve your Page Load Time though. Because in the end some bottlenecks might actually be impact multiple metrics at the same time, such as a large non-deferred JavaScript bundle which is impacting both First Input Delay and Page Load Time.

As a result, Page Load Time and for example DOM Content Loaded might be convenient for technical stakeholders. For example to get a sense of the performance hygiene of a domain of specific webpages and tracking its progress over time.

Other stakeholders

Average Page Load time should not be used by conversion minded stakeholders though. It isn't a good metric to get an idea of how webpages are experienced by real users, nor the impact on bounce.

Especially looking at just one metric -as I saw happening in a competitor's pagespeed audit I received- is very wrong. Pagespeed and the impact on bounce just is more nuanced. Pagespeed also isn't just a number, as no user nor their conditions are the same. Exactly the reason why the layout of PageSpeed Insights changed last November.

Money you might be leaving on the table

Maybe you were already using different data than just Google Analytics average page load time. If not, try to focus on metrics that actually has a correlation with bounce. Google's Core Web Vitals is a very good start already.

Want to find out if your webshop is leaving money on the table? You can check your own Core Web Vitals as experienced by real users and potential ROI on my website.