PageSpeed Insights, Google’s page-loading speed analysis tool. According to Google’s statement, page load speed will affect the ranking of the website in search results. In July 2018, this will also apply on mobile platforms. As a result, more and more people are confident in using PageSpeed Insights, Google’s page-loading speed analysis tool.
But the effectiveness of PageSpeed Insights is still a big question in the website development community. This article will explore how this tool works to answer the question of reliability, usefulness, and its impact on the search rankings of websites on Google.
How PageSpeed Insights works
From the first day being put into use, the algorithm of PageSpeed Insights has been almost unchanged so far. However, Google has changed the interface so that users can easily capture more information. Previously, the tool only displayed a single number representing the analysis results. Now, we have up to 2 assessments corresponding to the 2 analytical aspects of PageSpeed Insights:
Speed
Page speed or loading speed is calculated by Google from the moment the browser starts to load until the user sees a meaningful content on the website. This process, which is called First Contentful Paint (FCP), is the period of time from when the screen is empty until the content appears in writing or pictures (including background images). A fast page load speed will help keep users on the page longer, stimulate interaction and increase conversion rates.
Results of website speed analysis based on FCP and DCL
The page loading speed is evaluated by PageSpeed Insights through the Chrome User Experience Report (CrUX) function – a user experience report on Chrome. In addition, Chrome users must also disable sync encryption in order for CrUX to work.
The data collected by CrUX, in addition to FCP, also has First Paint (FP) – the time it takes the browser to load the first non-meaningful content of the page, the onload – the “download” function is performed when the resources Essential download has been completed.
In it, there is data that also contributes to determining page load speed: DOMContentLoaded (DCL). DCL is the amount of time it takes for the browser to load the most basic infrastructure component of a website – the HTML file. But this is just a plain HTML file, without any extra elements like stylesheets, images or subframes.
In summary, page load speed is analyzed by PageSpeed Insights by gathering FCP and DCL information from the real user experience of the Chrome browser user.
Optimization
For website optimization results, PageSpeed Insights does not retrieve real user data. Instead, PageSpeed Insights will scan and analyze the website based on its standards.
If your site does not reach the maximum score, PageSpeed Insights will suggest ways to fix, optimize performance. These suggestions, in theory, will improve page loading speed and website performance. They are technical optimization rules such as compressing CSS, Javascript and images, increasing server response time, or reducing the content size, etc.
How PageSpeed Insights works and the actual results
Many website designers have used PageSpeed Insights as a tool to measure website speed and performance. Not to mention, this is also a tool developed by Google, so more people trust it than other tools.
However, in the past and present, there are many doubts about the effectiveness of PageSpeed Insights.
Page load speed is a very difficult number to determine. It is influenced by many aspects and different factors. Developers can improve server quality, optimize website structure to help pages load faster. But actually, page load time mainly depends on the quality of the user-side connection and other factors such as: geographic location, cache (cache), browser type and device used …
Actual website loading speed also depends on the quality of the visitor’s connection
PageSpeed Insights’ collection method makes the data unreliable. CrUX collects information of any Chrome browser user who has this feature enabled. Thus, every Chrome user in the world who has ever visited your website will also impact the speed index in PageSpeed Insights.
Suppose that website visitors are divided into two groups. Group (1) has a small number but very good transmissions and equipment, somewhere very close to the web server and once visited should have cached. Thus, the page loading speed of group (1) will be extremely fast. But the group (2) is more crowded but lives at the other end of the Earth, uses dial-up network and has not accessed it at first, so there is no caching in the machine. Group (2) will wait a long time to load your site.
And yet, PageSpeed Insights gathers it all together to produce an average speed figure!
This does not reflect the experience of the majority of users accessing the website, nor does it speak of your optimal performance. In the example above, it is possible that PageSpeed Insights will produce high results but users in group (2) will find it slow to enter your site.
The above example is only a hypothesis. Errors can be large or not depending on the number and characteristics of users. But one thing is for sure: only through PageSpeed Insights, you cannot know how the majority of users have experienced the page loading experience on your website.
In addition, PageSpeed Insights does not give the amount of time needed to download components or the whole website. All you can know is the site’s FCP and DCL index. Maybe users need a little time to download meaningful content first on the website, but if the rest just loads forever, it cannot be called fast.
In addition, CrUX is a function unique to the Chrome browser. If users do not use Chrome, but instead Firefox or Internet Explorer to access the website, PageSpeed Insights will not be able to retrieve the data. You will only be able to consider the page loading speed of Chrome users. So, you cannot optimize the website comprehensively.
In addition to speed, PageSpeed Insights also analyzes and recommends page optimization according to Google rules. Technically speaking, following these guidelines will reduce page load time. However, in practice, following these suggestions is not always effective.
PageSpeed Insights seldom changes proposals based on actual situation. That is, in any website, whether the content is different, the structure is different, the tool often makes the same suggestions!
This is not because it is incorrect. The problem is that PageSpeed Insights can be a cause of slow website loading. No one knows where the real problem lies. So sometimes when you follow the 9/10 recommendations of PageSpeed Insights, the analysis results remain unchanged. Perhaps if all of the PageSpeed Insights suggestions have been completed, you may be able to improve their results.
However, this is often impossible!
PageSpeed Insights applies very rigid optimization rules, exemplary. This tool often suggests developers change the order of the code snippets of the page. However, if you follow, you will break the page structure, even paralyzing existing functions. In other cases, the actual page load speed is very fast, even under 1 second, but the PageSpeed Insights results are only average.
In addition, PageSpeed Insights also requires removal of effects for the website. These can be effects that increase the smoothness of movement or page change. However, websites that load quickly but are moving are hard and boring to attract users? You should consider trading these people-oriented elements in exchange for just a few milliseconds of page load speed.
In a nutshell, PageSpeed Insights does not understand that a slow loading website is not just because of uncompressed images or render-blocking resources, but there are many other factors.
The problem here is not that you do not optimize the website well, but rather the way of PageSpeed Insights.
How does PageSpeed Insights affect SEO?
As mentioned, page load speed is also a factor in Google’s search algorithm on desktop and mobile. Therefore, many SEO-er care and optimize for PageSpeed Insights. But how will the results measured from PageSpeed Insights affect how well the site ranks?
Does PageSpeed Insights affect SEO?
In 2013, moz.com – the world’s leading website design and SEO site – conducted a test of this. Moz checked 100,000 websites with 50 pages for each keyword. And the test results showed that: page load speed has little effect on the search ranking of the website.
The above results are true for both short keywords and long keywords (long-tail keyword). Generally, short keywords with a general meaning will be very competitive, so the search engine may consider the secondary factors when ranking, especially the first position. But Moz’s employees could not find any link between the speed of the page loading and the ranking of the website with the short keyword.
Instead, the results show that the factor affecting search rankings is Time to First Byte (TTFB). TTFB is the time it takes for a browser to load the first data byte of a website. In sites with the same keywords and optimized SEO, the shorter the TTFB, the higher the search ranking. This TTFB parameter is heavily dependent on parameters such as: transmission latency, quality of web server …
Everything shows that: optimizing your site for a fast loading speed does not necessarily give you a good place in Google’s search results. Therefore, 100/100 PageSpeed Insights points are not really worth it for you. Moreover, neither the PageSpeed Insights nor Google optimization rules are effective. Only suggestions related to improving the quality of infrastructure (server, line, connection flow …) will affect the search rankings of the website.
According to the logical thinking of the operating principle of the Google search engine, page load speed cannot have a big impact on SEO. It is still in Google’s algorithm, but cannot be given as much priority as “relevance” and “popularity”!
Pay attention to the conversion rate instead of speed
Understand the impact of page load speed on SEO in the following way:
Google will penalize the ranking of websites with a high bounce rate. And there are 2 factors that reduce page bounce rate: high conversion rate and fast page load speed.
According to statistics, users prefer websites with an average download speed of under 3 seconds.
So, if you optimize for a website that has a loading speed of approximately 3 seconds, it will keep users coming back, the bounce rate will be lower and Google will consider your website to be useful for users and improve ranking position.
That is the motivation that website and SEO designers need to optimize their website. But you do not need to try too much to reduce the download speed to 1 second or reach a maximum of 100/100 PageSpeed Insights. Your website loads like lightning but does not leave any impression to customers, they will also exit the page. Instead, pay more attention to conversion rates.
To achieve a high conversion rate, your website needs to have quality content, impressive interface, reasonable design and know how to stimulate the need for interaction from users. It is these things that are the leading factors for Google to determine the usefulness and order of websites on the search results page. Google also said in the July 2018 update: “[…] a slow-loading site can still rank highly if it has great and relevant content [to search keywords]”.
So what should this tool be used for?
By reading this, you will be wondering: PageSpeed Insights figures are so theoretical, so what does Google continue to grow?
In fact, PageSpeed Insights is still useful for analyzing errors when your website loads too slowly, has high bounce rates and low conversion rates. Using PageSpeed Insights, you will know your website has not yet met the optimal rules, so that it can be adjusted more appropriately.
But with the above analysis, you should not try to get the maximum score in PageSpeed Insights to do. This tool does not accurately reflect user experience and page load speed. It also does not help you improve page performance nor improve actual SEO performance.
PageSpeed Insights is only useful when used as a speed analysis tool for websites. Google may be wrong to operate PageSpeed Insights in the current way. Or maybe we used it the wrong way in the first place!