The answer to one of my most asked questions (spoiler alert: it's because of SEO and UX).
There are multiple websites and toolings that help you measure your perfomance score. Three of the best known might be Pingdom Tools, GTmetrix and Google's Pagespeed Insights (or maybe even it's underyling analysis engine: Lighthouse). When you are auditing your website, which one should you use?
GTmetrix announced that a GTmetrix analysis will be using Core Web Vitals as of the end of september. This means GTmetrix is safe to use when you also want to get an impression of the impact on UX.
Technical checklist vs realtime mobile data
I consider Pingdom Tools and GTmetrix as practical technical checklists (with some anti-patterns / nuance as well, but that's another story), where pagespeed metrics such as onload or DOMContentLoaded alone unfortunately aren't telling the whole UX story. Additionally, the default testing with Pingdom Tools and GTmetrix is done using a desktop device. Due to CPU boundaries/differences, this is not representative for mobile users.
- one could do a better job towards user engagement when using critical CSS, while DOMContentLoaded could still be bad, but at least UX is reasonable;
- or be doing a worse job when using spinner/splash screens, having a fast DOMContentLoaded, but users won't see anything meaningful soon, meaning no early user engagement.
That's where the UX minded metrics of PageSpeed Insights (and for example Lighthouse, Webpagetest.org) come in handy! Especially as Google's new Core Web Vitals is all about (technical) UX as well.
PageSpeed Insights raises questions
Within the field of SEO specialists and developers, there are quite some questions. Even after publishing this post, someone introduced me into a (Dutch) LinkedIn discussion regarding this exact same topic: should you use PageSpeed Insights or GTmetrix when being an SEO specialist or (Wordpress) developer?
The following topics were addressed:
1. "Are the scores of a speed test (in this case Google PageSpeed Insights) important or is the overall speed important?"
The answer to this question depends on what you consider being the overall speed. The person dropping this question, might have meant the overall experience instead of overall speed.
Some might still think loading speed can be expressed into one single metric. However, towards real user experience, this isn't the case. This was never the case, but using Lighthouse Google just found a good way to visualize this.
As a developer, you could trigger an early user engagement by providing an optimal perceived performance, while building the rest of the webpage or loading other assets in the background.
However, the user shouldn't be disturbed, otherwise they might loose focus when they already started reading and might leave your website or webshop as they got frustrated. The CLS metric, part of Google's Core Web Vitals, is an example of a non-pagespeed related metric, although impact UX and thus conversion in real life.
So, to answer this question: the overall scores is just an overall indication. To get a grasp of possible UX impact, you should use PageSpeed Insights and look into the individual metrics. These metrics and insights aren't provided by GTmetrix.
2. "PageSpeed does not determine your ranking"
Although your PageSpeed Insights result does not impact your score directly, it gives insights into how it could impact real user experience, as described above. Three out of six metrics are part of the Core Web Vitals, which will become a ranking factor in 2021, as announced by Google.
So, to answer this question: yes it does, but indirectly and based on real user experience data also known as CrUX. Obviously, content is still king! Using PageSpeed Insights might inform you in a non-technical way if you are passing the Core Web Vitals assessment and thus could benefit when competing with your online competitors.
3. "So a split second difference does nothing at all"
As Core Web Vitals is working with thresholds, you might want to end up in the green buckets for each metric from a ranking as well as conversion perspective. And with Core Web Vitals, we do not only have a reason to focus on pagespeed, but we also have clear upper limits to achieve.
So, to answer this question: yes it could matter when looking at the Core Web Vitals thresholds as illustrated within a PageSpeed Insights test.
4. "But PageSpeed Insights score is fluctuating, even from minute to minute"
This is correct unfortunately, but the same applies to real user experience. And as PageSpeed Insights tests are done using real devices (being a Motorola Moto G4), it is simulating real device behaviour.
In real life, your user's device could perform differently from time to time, impacting real user experience. Even if everyone would use the same device.
The outside temperature and thus temperature of the device will play a role, how many programs are running on a device, and it the device throttling because of a near empty battery?
So if the score fluctuates, it means that there is something in the HTML or JS code that can cause those fluctuations or diversity in real user experience. Something that you won't find out by using solely using GTmetrix.
🚀 Happy (online) speeding!