Why You’ll Never Get a 100 Google Pagespeed Insights Score

I’m a woman obsessed. So once I decided I wanted to get that perfect 100 score from Google PageSpeed Insights, I went for it and wouldn’t. Let. Go.

I challenged myself to make it happen with a larger site because, well, you gotta just go for it sometimes! I figured it would be easy given all the tools I had at my disposal, including WP CheckupHummingbird and WP Smush Pro.

I can definitely say with no confidence at all that I achieved my goal. (FYI: That’s my way of making myself feel better.)

Today, I’ll explain what I learned instead: Every site can’t get a perfect score and Google PageSpeed Insights isn’t actually the best benchmark for site speed, but it’s still worth using.

I’ll also go over what I did to improve my site’s performance, I’ll reveal my shocking score, why your site’s speed is still important, and a reasonable score you should try to achieve.

Why Site Speed Is Important

Why was I so obsessed with my site’s page speed? Because page speed has become an ever increasing factor in getting visitors to stay on your site.

In 2009, Google conducted an experiment and found they lose 20% of their traffic for every 100ms of load time for a page. In the same year, studies by Akamai and Gomez.com found that 47% of your visitors will abandon your site after 10 seconds or less.

In 2016, the numbers became more shocking.

Statistics compiled by Hosting Facts found:

  • Site abandonment was 40% if the loading time is more than three seconds. (eight seconds less in 2016 than compared to 2009.)
  • Site speed also affects your bottom line since 51% of online US shoppers say it’s the top reason for abandoning their shopping cart.
  • An increase in site speed from eight to two seconds can boost your conversion rate by 74%.

This trend is also set to continue in 2017 where there’s expected to be more internet traffic than all prior years combined.

Maile Ohye, Google’s Developer Program’s Tech Lead, noted in her video Site Performance For Webmasters that your site should load in 2 seconds or less, but those at Google aim for half a second, or 500 milliseconds.

Ohye also talks about a study Microsoft and Google did in 2010 where they purposefully delayed site loading times. At 500 ms, customer dissatisfaction went up by 0.9% and when loading times climbed to two seconds, dissatisfaction also went up by 3.8%.

This may not seem like much, but visitors were so angry, Microsoft quickly pulled the experiment.

Bottom line: If you want people to visit your site, stay longer and actually come back, a major factor in achieving this is a fast page load time.

What is Google PageSpeed Insights

Google PageSpeed Insights is a tool you can use to rate your site’s performance with a score ranging from zero to 100 for both mobile and desktop versions.

It checks for ways your site could improve its performance for the time it takes to load above-the-fold content and the time it takes for the whole page to load.

The higher you score, the better your site is measured to perform.

My Point-by-Point Play…

Regardless of why I needed that 100 score, it gleamed in my mind like a first place trophy and I had to have it on my mantle! I was willing to pull out all the stops to get it so here’s what I did to improve my site’s performance before steam started blowing out both my ears.

Feel free to follow along and create your own exact replica test site so you can see my frustration progress first-hand.

After I created and installed a fresh WordPress Multisite network, I added these plugins:

Akismet BuddyPress Cloner Contact Form 7 CoursePress Pro Disqus Duplicate Post Hello Dolly Hustle MarketPress Membership 2 Pro Multisite Privacy Smartcrawl Snapshot Pro Tickera Defender Hummingbird WP Smush Pro WPMU DEV Dashboard YARPP

The only two plugins I didn’t network activate were Hummingbird and WP Smush Pro since I was going to configure them later to optimize my test site. I also configured Defender, Tickera and I set up Multisite Privacy to protect the site’s SEO while I performed this experiment.

If you don’t currently have a WPMU DEV membership to use many of the plugins in the list above, I highly recommend signing up for a free trial, which will give you access to everything.

The themes I installed were:

I also created an identical post and page with 25 paragraphs (2095 words) of generated Lorem Ipsum text. Inside each of them, I added four images in their original size:

Next, I used the Duplicate Post plugin I installed to bulk copy the post and page I created. I deleted the “Hello World” post and copied the one I created in small batches until I had a total of 1000 posts.

Then, I copied the post I created so I ended up with 50 total pages. In the end, the images I added were also cloned and I ended up with 144 unoptimized images.

Benchmark Test

Now that my fairly large test site was ready, I needed to enter the domain into the Google PageSpeed Insights tool so I can see what score I get before optimizing the site. That way, I could compare my test site scores from before and after.

I entered my domain in the text field, then clicked on Analyze. My results were 35/100 for mobile and 37/100 for desktop.

Ouch!

Google PageSpeed Insights scoring.
Needless to say, my large test site did not perform well.

I also checked my test site with WP Checkup to see where I could make specific changes to improve the performance of my site since Google PageSpeed Insights doesn’t offer these details. That way, I could take a lot of the guesswork out of the optimization process.

I entered my domain name into the text field, then clicked on the Free Scan button.

WP Checkup page.
You can check your site’s performance with WP Checkup.

I got an overall score of 68/100. That’s not that bad considering my test site was highly unoptimized, but it was only that high because I was using Defender and I got a 100 score for security as a result.

On the other hand, my score for performance was 37/100 and I got 67/100 for SEO.

WP Checkup performance result.
My test site’s score in WP Checkup was terrible for performance.

I also got to see where I needed to make improvements and I clicked the How to Fix buttons to get details on how to, er, fix the performance issues on my site.

Finally, I benchmarked my test site by using Pingdom’s Website Speed Test to check its speed on the other side of the world compared to where my server is located. Since Google PageSpeed Insights can’t thoroughly check your site’s speed, Pingdom can help fill in this gap.

I entered my domain into the text box, chose a location from the drop down box, then clicked on Start Test. I got a score of 77/100 and my site loaded in 5.97 seconds. Eek!

Pingdom score.
You can accurately test your site’s speed in different locations with Pingdom.

Now that I could see where I needed to improve, I could start going down the performance mending rabbit hole.

Optimizing Images

To start out, if you’re following along with me, network activate the Hummingbird and WP Smush Pro plugins, then go to Settings > WP Smush to adjust some of the image options.

You can leave the first two options enabled and also enabled the Resize original imagesSuper-smush my imagesInclude my original full-size images and Convert PNG to JPEG (lossy) options by clicking the toggle buttons next to the setting names on the list.

WP Smush Pro settings page.
Enabling Super Smush saves a lot of room and helps increase your page speed.

Then, save your changes by clicking the Update Settings button at the bottom of the page.

Once that’s done, head over to your test site and go to Media > WP Smush. Click the Bulk Smush button to start smushing your images.

Then, sit back and relax for a few minutes while your images are compressed.

Bulk smushing images.
You can compress all your images in bulk and smush newly uploaded ones automatically.

Browser Caching, Minification and More

Now that your images are compressed and optimized, it’s time to optimize your entire site with Hummingbird. Go to Hummingbird > Dashboard in your admin dashboard and under Minification, select the Only Super Admins can minify option in the drop down box.

Next, click the Start Caching My Website button to turn on browser caching. Now that browser caching is enabled, go back to Hummingbird > Dashboard and click the Compress My Website button.

Hummingbird options.
Get your site optimized in a few click in Hummingbird’s Dashboard page.

Then, go back to the dashboard page one more time and scroll down to the bottom of the page to view the CloudFlare settings. Take a moment to set up

Take a moment to set up CloudFlare for your site, then come back to the Hummingbird settings and enter your CloudFlare email and Global API key. You can click the Need help getting your API Key? link at the bottom of the form to reveal details on where to get this information.

I chose Cloudflare’s free account type, but you can upgrade to a premium plan if you want.

Hummingbird's Cloudflare settings.
Enter your Cloudflare email and Global API key to let them work as a team.

Click Connect and you’re all set to run your first performance scan to see your progress so far. Go to Hummingbird > Performance Report and click the Test My Website button to start the scan.

Safely Minifying Files

After the scan has completed, click the Improve Score button under any result listed that isn’t at 100 to fix it. Try clicking that button for Minify Javascript, then on Minify with WP Hummingbird.

On the next page, click Check Files to scan your site for the files in your site that can be minified.

Keep in mind that themes and plugins that don’t follow the standard structure could cause some CSS and Javascript files to not be listed. Hummingbird would still detect them as needing to be minified, but won’t be able to minify it.

This isn’t always something you can change and it leads to a lower score, even though it won’t drastically affect your site speed.

This is one of the reasons why trying to aim for a perfect score in Google PageSpeed Insights, Hummingbird and WP Checkup isn’t always possible and you don’t need to feel pressured to try and achieve it.

When you’re minifying and combining files in Hummingbird, especially to remove render-blocking issues, there’s also a critical detail you should also be aware of and it’s displayed toward the top of the page:

“Moving files between the header and footer of your page can break your website. We recommend tweaking and checking each file as you go and if a setting causes errors then revert the setting here.”

Some files can’t be moved to the footer no matter what you do since the theme or plugin that includes the file requires it to be left where it is inherently on your site or above the fold.

If you move one of these immovable files, there’s a strong chance it will break your site. This is yet another reason why you won’t always be able to reach a perfect 100 score in Google PageSpeed Insights, Hummingbird and WP Checkup.

This is the case with some of the plugins I have installed for my test site, meaning I won’t ever be able to achieve a perfect 100 score.

But that doesn’t mean my site won’t be fast.

You can choose to include, minify and combine files in bulk, then choose a position for each file one-by-one between saving and check that your site is still working before positioning another file.

Minification settings.
It’s recommended that you position your files manually instead of in bulk.

It’s also important to note that if you decide you want to use another plugin with similar features to Hummingbird, you shouldn’t enable the same kind of options in both plugins since they would interfere with each other and break your site. Instead, configure matching options in one, but not the other.

Quick tip: When you’re minifying files, turn off caching in Hummingbird and any other plugins you’re using with this option. Since caching temporarily saves your site so visitors don’t have to load content every time they visit, it means you won’t see any changes to your site that you make to improve performance until you disable it. 

Once you’re done, you can also set up more options like uptime and blacklist monitoring if you want and details on how to do this can be found in Hummingbird’s Usage documentation.

Final Results

After realizing there wasn’t much I could do to reach that 100 score with the plugins and themes I chose, I decided to take one last test of my site and call it a day.

My results were shocking and not because I actually got close to that elusive 100 score, but because they were still so terrible and there’s not a whole lot more I can do about it unless I want to make major changes to my site.

In Google PageSpeed Insights, I scored 48 for mobile and 50 for desktop.

Final PageSpeed Insights score
My rather lackluster final score.

To improve my score, I need to resize every single image on my site so they’re not so big. Doable, but that would take a long time and what if I need my images that large? Similarly, photographers generally aren’t going to want to discount their work by reducing the quality and size.

I also have too many plugins installed for the size of my server. Unless I want to start eliminating plugins or upgrade my hosting plan, there isn’t anything I can do here, either.

While on this test site I could eliminate most of the plugins I installed, on a live site, each plugin has a purpose and while it could be possible to delete some, it may be difficult to eliminate most of them.

Similarly, in WP Checkup, I scored 72 overall, 50 for performance, 67 for SEO and 100 for security.

Final WP Checkup score
My final WP Checkup results were just as disappointing.

After reviewing my WP Checkup score, the story was the same, but in more detail. There were options to fix the issues, but there’s only so much a person can do sometimes.

Finally, in Pingdom, my score was a lot higher and my test site turned out to be faster to load at 1.32 seconds.

Final Pingdom results
My site ended up being pretty fast, surprisingly.

Not too bad for a site that was a tremendous flop in its performance scores. This last result says everything about why a perfect 100 Google PageSpeed Insights (and WP Checkup) score isn’t always achievable and also not really a problem in most cases.

Following the Google Brick Road…

It’s common to associate a high score with a blazing fast site, but Google isn’t perfectly accurate at determining your site’s speed according to Google’s documentation:

However, since the performance of a network connection varies considerably, PageSpeed Insights only considers the network-independent aspects of page performance: the server configuration, the HTML structure of a page, and its use of external resources such as images, JavaScript, and CSS. Implementing the suggestions should improve the relative performance of the page. However, the absolute performance of the page will still be dependent upon a user’s network connection.

This means that while it’s an infuriating excellent tool you can use to measure your site’s overall performance, it doesn’t accurately measure your site’s speed when it comes to how your visitors perceive it around the globe.

While it’s a tool that can be used to help you pinpoint where you can improve your site’s performance, it shouldn’t be seen as the only determining factor for whether your site performs well because that’s not what it was designed to do.

You can check out details on other tools you can use to rate your site’s performance in the post How Fast Is Your WordPress Site? Find out with These Free Speed Testing Tools.

Is the Elusive 100 Score Even Possible?

Yes and no.

Confused yet?

The answer isn’t as cut and dry as one would hope. If you did a quick Google search, you would find many articles detailing how you can achieve a Google PageSpeed Insights score of 100 and fairly easily to boot.

So case closed, right? Not quite.

With all the articles out there saying a score of 100 is possible, I even started to believe it until I looked into it more critically and found their test sites are basic and so much so, the sites are practically static.

Sites that are static – and have some text, a few images and not much else – are bound to load quickly after making some basic performance enhancements. Overall. a site like this would earn a higher score without much effort.

WordPress is inherently dynamic which automatically makes it more difficult to achieve a PageSpeed Insights score of 100. While it’s still possible if your WordPress site is as basic as possible, that’s not always practical.

Sites that expect to display tons content are going to have a more difficult time achieving a higher score. Dynamic content such as videos, accordion tabs, forms, live maps and the copious features and capabilities plugins offer are all going to translate into an increased strain on your server and site speed.

Your Google PageSpeed Insights score is expected to be lower as a result.

A fresh install of WordPress with no content and the Twenty Sixteen theme installed yields a score of 71/100 for mobile and 85/100 for the desktop version. While these scores may vary slightly depending on your host, it illustrates the point that a vanilla WordPress site still doesn’t get a perfect score.

Score for a fresh install of WordPress.
A brand new installation of WordPress doesn’t achieve a 100 score.

That being said, Google’s documentation on the tool also states that a site which scores 85/100 and above is considered to be running well. If your site is reaching that score, getting a perfect grade is a moot point.

It doesn’t mean you should stop trying to improve your site’s performance altogether, it just means that your site can be blazing fast, perform extremely well and also not score 100 in Google PageSpeed Insights.

Above all, you shouldn’t be obsessed over a perfect score like me and focus your attention on the UX and UI of your site. If your visitors get a fast site that’s easy to use, then you’re golden.

Making Your Site a Performance Power House

I was able to achieve a much faster page speed even though my Google PageSpeed Insights score was atrocious, even after making so many improvements.

At the end of the day, you may not be able to reach that elusive perfect performance score and that’s perfectly alright if your site is fast and it’s easy to use. You don’t need a 100 score to have a well-performing site.

Jenni McKinnon
Are you as obsessed as I was with trying to get a perfect 100 score on Google PageSpeed Insights? Are you still trying to achieve a perfect score? What challenges have you come across trying to achieve a higher-performing site? I want to know so share your experience in the comments below.

23 Responses

  • New Recruit

    Kinda strange article :)

    I charge my clients $300-400 to get 100/100/100 scores for a WordPress site. Yes, from time to time it can happen, that achieving all 100 will need some big changes to site structure. But still, achieving this goal is quite possible.

    Lazy load heavy content, use of critical css, + all the other rules from Insights do the job.

    • Hey,

      That’s true and that’s why I mentioned it in the article.

      I wrote this article because a lot of people are obsessed with getting a perfect score when they’re still scoring high and they’re site works perfectly and has a great UX and UI. At that point, there’s no need to aim for a perfect score at the risk of reducing the site’s quality. I didn’t want people to be unnecessarily worried about their scores, thinking it was the end of the world if they didn’t get 100.

      Cheers,

      Jenni

  • Flash Drive

    I’ve been doing SEO since March of 1995 (yeah, seriously). I can say it’s difficult for average person to get a 97+ on Pagespeed Insights, but it’s very very possible; it just cannot be done with one-click plugins and the presumption that you can leave your site structure as-is.

    For example
    a) optimization of wp-config.php
    b) optimization of htaccess for caching/TTY/expiration headers
    c) manual minification of main CSS and JS files
    d) moving CSS to footer and putting ‘critical css” in header
    e) merging “dozen JS” files into one or three vs ten, then minify
    f) remove useless elements like emoticons, or comment css; consider loading jquery via Google CDN
    g) use web-safe font stacks, or preload Google font CSS in your own CSS style sheet to direct load the fonts, not the CSS which THEN loads the fonts
    h) custom functions to remove clutter in head, remove versioning, de-enque CSS from plugins; disable CSS not used at all, reset default image compression
    i) trim your theme functions to remove loading elements like BBpress, WooCommerce if not actually using that stuff

    Downside is that your theme will be “locked” at the version you’re at, unless you redo the manual modifications — and you don’t want to use child theme as that is a redirect from get go.

    Additionally, technologies like PHP7, mod_pagespeed, http/2 running on server side have huge impact on speed.

    Our shtml based sites have perfect 100/100/100; WordPress 99/100 and 100/100.

    It is doable, just not “out of the box.”

    :-)

    • Hey,

      You’re right, and as I mentioned in the post, it’s possible, it’s just not going to be the best move for everyone and in particular those with sites that are already blazing fast and have an excellent UX and UI. In such cases, there’s no need to be obsessed with gaining that perfect score.

      That’s why I wrote the article. I wanted to make sure people knew that your site can be great, but still not score a perfect 100 and that this was okay in many cases.

      Cheers,

      Jenni

  • WordPress Enthusiast

    The bits about pagespeed indicator that anoy me the most is things like punishment for external scripts not being cached long enough like google analytics or such things found in video embeds, and the huge amount of posts for compressing some images, its quite disproportionate compared to the different it makes to the humans looking at the website. Or also when it gets annoyed about the wprocket cache files.

    Thing is there is nothing I can do about this for example

    Enable compression for the following resources to reduce their transfer size by 66.4KiB (72% reduction).

    Compressing https://cdn.rlets.com/capture_static/mms/mms.js? could save 60.7KiB (72% reduction).
    Compressing https://rtsys.rtrk.com.au/rct_lct/js/rlrct1.js could save 5.7KiB (71% reduction).

    other than take them out or hide them . The providers of the scripts wont put compression on them for me.

    Do I just clone those onto the site but then if the files are updated i have to manually update them.

    • Support Gorilla

      Hi Jeremy!

      I agree that this is kind of inconsistency within testing algorithms and I think it is affection more tools like e.g. GMetrix too, not only the one from Google. You are right however about the solution: you are not able to make changes (like turning on compression or doing any kind of image optimization) to resources that you don’t have access to so getting them on your server seems to be only reasonable way to deal with them. You could probably go for more complex solution like setting up some sort of your own “caching proxy” to serve such resources and apply “optimization” there but I believe that may turn out to be way to complex to justify the cost :) I’m not sure either about how search engines would treat that, I suppose that could lower your site’s ranks. The choice then is either to just accept it “as is” or, as you already mentioned, clone such resources and take care of keeping them up to date manually.

      Best regards,
      Adam

  • Dear “obsessed woman”, dear Jenny, I’m sorry for your frustration, and I understand the reasons you adopted to state “Every site can’t get a perfect score and Google PageSpeed Insights isn’t actually the best benchmark for site speed, but it’s still worth using”. I read your article just as published, I wanted to reply but since then I was really busy by my job and my new job, for which I was setting up this site So I’m doing now.

    I agree with you on the fact that getting 100/100 on Google PageSpeed Insights and all the other metrix tool is tough.
    I don’t agree with you that “Get a 100 Score on Google PageSpeed Insights for WordPress Is Never Going to Happen”.
    I agree to the fact that you don’t need to get obsessed by reaching the highest scores on all the metrix tools.
    I seriously affirm that if you think at the reason why these metrix tools exist, and what is the final target of achieving the best scores, well it is best to consider to reach them.
    All the metrix tools are measuring in numbers the user experience. So if a site is slow, then the user is not happy waiting for the page to load, if it is a common site. If the site is important, as a certain authority, or represent a well known brand, or a celebrity, well the visitor will wait a bit to read about what she is already interested in. The owners of this kind of sites continuously buy “user experience” by making their brands great, spending in every kind of advertising media, or simply cause they are famous, so the media spend money talking about them. Obvious observation.
    I analyzed a lot of well know WordPress sites, and it’s rare find one of them perfectly optimized for speed.

    I optimized a bunch of WordPress sites, and Google 100/100 is possible and I prove it in my site.

    You said you created a fresh WordPress Multisite network, so first off I tried to understand if the hosting server you used may have some problems.
    It’s on , 8 other domains hosted on your IP address, created on 2016/10/22, located in New York, New York, United States (US). I can’t guess the plan, but their server are fast, and even if you pickup the smallest package, is still a decent environment. I exclude the server as cause of not perfect scoring.

    To me, the most significative improvement you show are the ones you read with Pingdom Website Speed Test, from 5.97s (Dec 13 at 17.18) to 1.32s (Dec 13 at 22.39).
    But the first test has been done from Melbourne, the second from New Your City. And testing from different locations switching continent can have a bigger impact on page load times. I’m sure you did not notice that, sometimes you chose a server, and the test has been done from another, it happens.
    I actually repeated the tests on your site, cause this is an intriguing topic to me, and I will like to base my future doing this work, I mean selling WordPress optimization and speed performance packages. So I am an obsessed man too.

    You may have done some edits on it, cause now the scores are different.
    Google PageSpeed Insights Mobile 48/100 Desktop 49/100
    Pingdom Website Speed Test B85 1.23s 505KB 14 Requests from New York City
    B85 4.02s 505KB 14 Requests from Melbourne
    B81 1.17s 505KB 15 Requests from New York City testing the address.
    You made your test on the version, not on , this diminish your scores of a couple of points, due to the added redirect.
    GTmetrix PageSpeed A99% YSlow A92% 2.2s 384 KB 13 Requests from Vancouver, Canada
    webpagetest.org First View 3.017s load time, 1.448s First Byte, 437KB 15 Requests from California USA – EC2 – Chrome – Cable
    First View 2.870s load time, 1.987s First Byte, 437KB 15 Requests from New York, NY USA – Chrome – Cable

    I don’t want to compare all these results with the ones you published, it’s too evident the WordPress installation environment changed from that day.
    But I’d like to make a couple of observations.
    – A page like this, with so little size and little requests must load much faster.
    – Google “Reduce server response time” warning reporting server responded in 0.80 seconds on mobile and 1.00 second on desktop, together with the too highest ttft, First Byte times measured by webpagetest.org, shows that no static cached pages are being served.
    – There are 4 not optimized images above the fold, with 91% of reduction, strange.
    – “Eliminate render-blocking JavaScript and CSS in above-the-fold content” can’t be fixed by Hummingbird plugin moving the combined and minified file at the end of the html code, cause there is no Critical Css inline, necessary to render the page.
    – You can try to use Google’s ModPageSpeed but I see it to many times not working in a perfect manner, as Google clearly affirms.
    – I agree with dimasmagadan, neotrope and Hrishikesh. 100/100 score can be achieved only if inline Critical CSS is used. The site of Hrishikesh is a well done example of a WordPress full optimization job, its homepage uses only the inline Critical CSS code to render all the page, except for He has been lucky enough that autoptimize and above the fold plugin worked. Autoptimize used to inject the Critical CSS has a limit on the size of the injected inline code. And its cache never purges, so in little time it become too big and requires manual deletion of files. Above the fold JS used to defer css, tends too many times to be caught itself as render-blocking JS, but in this case, since there is no external css file to load, it passes the test. This is a rare case, based on the jobs I’ve done.
    – There is no need to reduce the site quality, I learned how to score 100/100 with no layout changes. Usually on WP you need to leave out jquery from combination, and deferring it without moving to footer.
    – I think that in the near future the attention to the site speed will change, if Google will continue acting as now. I have been able to make free wordpress.com sites faster than the others, with manual optimization, cause no plugins can be used there, and made some simple tests to rank some well SEO written posts: Google acknowledged that in very little time making them ranking in first page, competing with high Authority websites. So WordPress speed optimization matters also from a SEO perspective.
    – I think that big site like the one we are writing on will make some moves in the near future too. Some numbers?
    The actual page I’ commenting on (size 1.6MB, Requests 127) load time from Melbourne 7.39s Grade B83, from New York City 2.77, Grade C73
    ( size 1.6MB, Requests 101) from Melbourne 5.29s, Grade B89, from New York City 1.94s, Grade B89
    (size 7.2MB, Requests 75) from Melbourne 4.95s, Grade C76, from New York City 2.11s, Grade D69.
    Not so good numbers.
    And if we give a look to ttfb, time to first byte using webpagetest.org…
    this page measures First Byte 1.181s
    the home page First Byte 0.638s
    and 0.954s
    all tested from: California USA – EC2 – Chrome – Cable Tester: i-5cc8e8e9
    Too high values, the recommended is under 0.3s. I know for sure that once a cache plugin serving static pages is active, this value drops down under 0.3s. Still they remain not so good numbers, the server environment is excellent, HTTP/2 compliant, CloudFlare CDN, it serves millions of blog so it is a very high end server. We can see a similar scenario on the sites hosted on wordpress.com servers. WordPress is a great platform, still it is a slow big giant. Needs to be optimizied for performance and speed.
    – A final consideration on Critical CSS. Everybody, between the experts I mean, talks about it. Google recommends using it. And there are a lot of tools to generate it. But, for what is my experience no tool is able to generate a working Critical CSS for WordPress. WordPress as a normal installation, with its amazing themes, lots of plugins, eCommerce enabled, the reason why people chooses WordPress. If we give a look to the various performance tests, the most of them are made on basic installation, using basic themes, just to let things easier, maybe on a dedicated VPS. And with no mention to Critical CSS. We here are talking about giving the best performances possible to real WordPress sites. Without saying to the owner that she needs to quit using this or that (there are some exception, though, some plugins out there are too heavy and slow down everything, wp-admin too). That been said, I well understood the reason the automation of Critical CSS fails, and on this little secret of mine I created a standardized method to do it, manually, tough. In a real website the CSSOM processed by the browser takes care of parsing all the assets and renders the site as the web dev wanted. When everything is condensed in a unique file, inline, the things gets complicated. A single rule in the wrong order could mess up the page layout. We must serve a perfect CSS code, otherwise the rendering fails. Is this complicated? Yes, of course. No user has to do or knows what the browser does to render the page she is viewing. WordPress is amazing thanks to its versatility. Yes. And today a WordPress site must be optimized. To me this will be a very hot topic in the years to come, and I am personally working to be one of the actors, serving WordPress site owners optimize and speed up their sites.

    Thanks Jenny for inspiring once again. I really appreciate your work, and hope my message will be useful.
    A curiosity: The worst site scoring ever seen? A WordPress site on BlueHost, Google PageSpeed Insights mobile and desktop 8/100.

    Have a nice time!
    Paul, from Italy

Comments are closed.