Optimize WordPress to get 100_100 Google PageSpeed, Grade A 100 GTmetrix, Grade A Pingdom Website Speed Test, Grade A Web Page Performance Test

Optimize WordPress to get 100/100 Google PageSpeed, Grade A 100 GTmetrix, Grade A Pingdom Website Speed Test, Grade A Web Page Performance Test


Optimize WordPress to get:
100/100 Google PageSpeed
Grade A 100 GTmetrix
Grade A Pingdom Website Speed Test
Grade A Web Page Performance Test
Best scores with other metrix tools
and serve your WordPress site as faster as its server can do

You are about to read…
  • If is it possible to optimize every WordPress site in order to make it faster as its server can deliver and how.
  • How hard is to achieve this target.
  • What the experts are saying about optimizing WordPress and how they deal with it.
  • What the experts think about maximum metrix scores achievement and their advice on how not to get mad at it.
  • Why a fast site is important and why a blazing fast above-the-fold content is the real target.
  • How we are able to optimize every WordPress site and make it faster as its server can deliver.
  • Why is very hard to forge a working WordPress optimization technique.
  • Why it does not exist a shiny plugins that makes all the job.
  • What we can do for Your WordPress site.
(maybe not in this order, follow along through the lines…)

The best and unique system to get the maximum of Google PageSpeed Score, GTmetrix, Pingdom Website Speed Test, Web Page Performance Test, and all of the other metrix tools, for your WordPress site. In other words, the best and unique system to make your WordPress site as faster as possible, without changing hosting or adding CDN.

Working on low budget shared hosting services, in a normal WordPress installation with a decent hosting provider will give great results. On VPS your site will load as faster as servers can be.

WordPress is a great platform, but sometimes it shows as a giant slow monster in terms of speed and overall performance. And every shiny plugin can slow it down a bit more, in despite of what it promises. Don’t be afraid of slow down side effects. Use WordPress as you always dreamt to, as faster as it really can be. Put our experts on work, let us to the hard job and enjoy your faster WordPress site ever.

Before start thinking on how to make your WordPress site faster, to start the journey of the long time used on learning, applying, and testing to improve it, maybe by switching hosting provider, moving to a VPS or dedicated server, using CDN, using performance plugins, think about all the time these tasks require. Now think about your WordPress site, you want to use it for your primary purpose, you don’t want to be consumed by its performance needs. You can hire a team of WordPress optimization experts that will do the hard job for you. Of course there will come the time to change hosting provider or hosting plan, maybe when your site will become popular and will need to handle hundreds of hourly visitors. Our settings will help you become popular faster, because your site will be high performance since its beginning. No worries about traffic spikes. It will handle with ease because it will be already optimized.

HOW WE DO THE JOB

LET’S START FROM AN EXAMPLE OF WHAT WE HAVE DONE

   Simple version

We have elaborated a unique system to make WordPress faster as its server can be, starting making tests on slow performance low budget shared hosting services, the worst situation in which WordPress could be installed, and on a WordPress with complex theme and lots of plugins installed (~150) and active (~75). By manually creating Critical CSS file and manually optimizing WordPress using a special combination of plugins (the most free) we are able to do what we proudly offer. All this process took us months of research, tests, verification of reproducibility and now we are able to process any kind of WordPress installation and give 100% successful results. The tasks are very complex, and took from 20 to 40-hour job per site, 2 different sections, to make it done.

   Tech version

WordPress installation multisite with complex Theme, full responsible, a lot of plugins, HST, full https access, Let’s Encrypt SSL, no CDN.
Test made on purpose on low budget shared hosting service, with not so perfect configuration, Intel(R) Xeon(R) CPU E3-1230 v3 @ 3.30GHz, 16 GB RAM, SSHD, Ngnix + Apache 2.4.23 + PHP 5.6.23, no keep-alive on server side at the time of the first test, then hoster switched it on, no kind of cache on server side: Opcode Cache, Memcached Cache, Redis, Varnish Cache.
Very high TTFB, over 700 ms, while processing WP database requests. TTFB under 300 ms for static files. A simple modding to Firefox browser with max persistent connections >10 will simply pull down php database, preventing access to WordPress admin back-end, triggers server warnings, and in a couple of minutes, black list the IP address of the connected server. A disaster.
Responses with no optimization: over 6 seconds.
We used this test environment because if a decent result could be achieved here, everywhere else would be better and give best and excellent results. Affirmation confirmed working on good servers, page loads on average above 0.6 seconds (this number needs always to take into account the distance between the webserver and the test server, tests made from different continents can lead to higher results).

Strategy applied to reach decent results in terms of speed and the max score on Google PageSpeed, Pingdom Tools, GT Metrix, etc.

  • Elaboration of Critical CSS (also named Critical Path CSS or Critical Above-the-Fold CSS)
  • All the possible .htaccess (or Apache or Nginx) configs and optimizations to enhance server cache, gzip compression
  • A check to avoid .htaccess bad rules and minimize redirects (this happens very often, due to copy and paste of the public .htaccess tricks without knowing how .htaccess works)
  • Selective activation of JS and CSS assets
  • Selective activation of plugins
  • Minification and combination of JS and CSS files.
  • Parsing defer of JavaScript
  • Fonts optimization
  • Minification of inline CSS and JS
  • Minification of HTML
  • Images optimization
  • Cache to serve static files

All of this global optimization settings will reduce the number of requests to the server and the total size transferred per page, in order to have the best TTFB as possible, and the lowest total loading time too. All this let us pass the “use less plugins as possible” myth, having WordPress running with ~150 plugins installed and ~75 active.

Nothing new, but…

WHY OUR SYSTEM IS UNIQUE?

  • Elaboration of Critical CSS with the techniques and tools publicity available don’t work as expected. Google itself with its PageSpeed Modules for webserver, and specifically the Prioritize Critical CSS Risks section, warns about that, “this filter could potentially show unstyled content before it loads the page completely, or in certain cases it could break rendering completely”, so no guarantee about good results. We can’t be able to find a decent explanation and working solution on the web.
  • The concept of reducing all the CSS assets into one unique little file, although highly recommended, is a tough task, and the tools available don’t lead to a reliable result. Especially when working with WordPress, a platform in which the Theme primarily adds a lot of stylesheet files, and dynamically generated too, then the Composers, then the other plugins. And on the average if a nice 10% of all CSS rules are used, let’s stay happy. The rest if lost bandwidth and highest page loading times.
  • Talking about WordPress speed optimization in general, without the use of Critical CSS, the suggested or advertised plugins acclaimed to give wonderful results, do the job, partially. Once one tries live knows what we are talking about, if some improvement is reached, then you must stay happy on them.
  • The common suggestions on how to optimize WordPress speed are always the same: choose a best hosting, use cache, minimize assets, use CDN, choose a lighter theme. Either expensive solutions and/or impractical ones, changing the theme could make loose the identity of the website, for instance.
  • WP Rockets experts finally came up with a really nice explanation on how WordPress site owners do not have to care too much to Google PageSpeed score, the 100/100 is not the real target, while speed must be it. wpmudev says “Why Trying to Get 95+ on Google PageSpeed Insights for Your WordPress Site Will Drive You Mad!”. GTmetrix experts suggest to use the free plugin W3 Total Cache, give a detail explanation of the best setting to be use to get some results, and advice to use their premium services ($250-$300) to solve the above-the-fold problems that that plugin is not enable to fix. And many other experts tell similar stories about their considerations and attitude about this topic.
    We agree with the concept of speed being the must target, but reaching 100/100 do helps, a lot, in having a fast website too. Despite what the experts say.

We elaborated a complex technique, that can take to a reliable Critical CSS Path file generation, and to build a unique CSS Stylesheet files too, this last one is suggested only on well-established sites, where the owner has set that the layout won’t change in the future, otherwise the single file needs to be tweaked, in case of little adds, or rebuilt, in case of brand new layout. This is the reason why only great brands and established sites, have that configuration file active, because of their engineers they actually have always working on the site development and maintenance.

The technique is not only complex, but also very time consuming. There is no automatic magic button with one click solution. No. All the job starts from a deep analysis of the WordPress site, of the theme used, of the plugins installed and activated. This leads to the choice of how many different sessions the site has, in terms of layout, so that specific Critical CSS path are built. The use of some tools to elaborate some first raw files and then a lot of manual work to make every CSS rule work with the required selector.

JOB DELIVERY

On average a couple of Critical CSS generation requires a 20 hours task.

On average a unique single CSS file generation requires a 40 hours task, and includes a couple of Critical CSS generation too.

[ Unfortunately for us about the 10% of the jobs done dealt with particular WordPress installs setup that required a lot more of work, time and efforts. We take the risk of having the job done in spite of this. If your WordPress config will result hard to work, we will promptly inform you about that and the possible delays on task execution, but no extra hour job will be ever charged. ]

Then there is the necessary PLUGINS INSTALLATION AND SETUP.

All this job needs to be done on a clone installation of the site, in order to apply the working setup and go live in a safe way. No broken sites allowed. Unless of test sites.

A tedious research we made is about the WordPress Plugins to be used in order to reach the desired target.

There are a lot of plugins out there, free and paid versions, and no one is able to satisfy our target needs alone. So putting together the working combination is not easy, it requires a lot of tests, and it is very time consuming.

We invested time and efforts on researching the good combination. It’s not perfect, it’s working. Having more than one plugin to rely on that works in conjunction with others is not simple. We find out how to tweak each plugin in order to have the desired benefit. And settings can vary based on WordPress overall configuration. Most of the plugin published in this field are continuously updated, searching for the better performance, and brand new once appears on monthly bases. No one resulted been able to do all the job by itself.

DUAL SCENARIO

A. Critical CSS added to the existent WordPress install.

All the css files used, remain still in use.

Ideal for the website with various content, multiple layouts, and experimenting like webmasters, loving adding new layout variation, or for the one who vary a lot using their composers.

B. Single CSS added as Critical CSS to the existent WordPress install.

A unique CSS file is used to render the full site.

Ideal for websites with an established layout and brand, who don’t plan to make important changes frequently.

NOTE: This scenario will work depending on the size of the Single CSS file generated. If too big, HTML size becomes heavy enough to trigger a warning, cause the page requires additional network round trips to render the above-the-fold content. At this point the solution will be chosen between using a Critical CSS file in addition to the Single CSS file, reducing the overall size of the CSS assets or recurring to Scenario A. The choice will be done by observing visually how the site will load better. Sometimes the use of Critical CSS file + Single CSS file produce strange loading effects, flashing, sliding menus, sliding widgets bars. The metrics tools will love both the configurations while the real user will prefer the one smoother, and that will be our choice. Not every theme shows this behavior. We already know how to deal with it. It is not possible to estimate the final size of the Single CSS file before having it done. On average we produce usable files. Big file only happens when a site is built using too many different layouts, overusing the composer capabilities, maybe while the owner was too enthusiastic to use astonishing effects. And usually these kind of sites are not so good to see, too much confusion there. If you note big brands that invest a lot on their sites, tend to have simple layouts or use different sections in order to have a clear separation on the topics, and more efficient develop of the section layout. And often they use a very big large and very lite static picture in the above-the-fold content, together with little or no menu, just to display as faster as possible. Or they are a very big brand and they don’t really care if their followers/fans need to wait a couple of seconds of page loading time, because of their unique and exclusive content.

In both situations, global optimization settings will be applied, by minimizing and combining JS and CSS files, enabling parsing defer, optimizing Fonts, minifying inline CSS and JS, minifying HTML, in order to reduce the number of requests to the server and the total size transferred per page, serving cached static pages, in order to have the best TTFB possible, and the lowest total loading time too.

CDN can be setup as an optional request, as a final addition, the most of the working and stable offers are based on monthly paid plans, so it will be necessary to understand the website bandwidth needs before applying for the best offer/plan available.

We are also able to pass the cookie-free domain warning. But this requires further steps and interventions on the existing WordPress setup, so we can discuss on how to proceed and give a custom quote for this particular job.

Exclusive pre/after optimization professional report.

Each job is delivered with an exclusive pre/after optimization professional report, where our engineers will note details on the WordPress hosting server environment, and will break down into details the most important metrix numbers, with the screenshots of the pre/after optimization tests of Google PageSpeed Insights, GTmetrix, Pingdom Website Speed Test, WebPagetest with relative annotations.

FINAL CONSIDERATIONS

ABOUT SPEED. TARGET RESULTS.

GOOGLE PAGESPEEDS INSIGHTS

target score mobile 100/100, desktop 100/100. Every result above 95/100 is good. Sometimes is impossible to reach 100/100 mobile User Experience, usually due to “Size tap targets appropriately” that cannot be fixed, if for instance, the cause is the scroll to top button or the search bar. Leave it there and don’t care about it.

While Google PageSpeeds scores near 100/100 matters, that’s not the same on the other metrix tools.

PINGDOM WEBSITE SPEED TEST

is more accommodating, it gives easily the highest scores, so to us while using this tool we give more attention at the Load time, choosing a server tester near the target one, and at the Performance insights, that may display important suggestions. Grade A100 is an easy target.

GTMETRIX

is a very detailed tool. With the PageSpeed score it is possible to get near to 100, but everything above 90 is good. With the YSlow score it is not possible to pass 96 without using a CDN. So near 90 without CDN and near 100 with CDN is the target.

Little details can be discarded here:

the PageSpeed algorithm tool gives too much attention to saving resources by the number of affected files rather than the entity of the saved bits. It activates the recommendation message even if the saved resource is about a couple of bits, regarding of the size of the inspected object. So don’t get mad at this. [If you are curious on how they get their reference numbers, they apply the minifications engine of W3 Total Cache, and the image optimization engine of WP Smush]

And, as a last consideration, the most important parameter here is as usual the Page Load Time. It must be taken into account the tester server and the target server location. If there is a continent between, nothing can make the entire website loading faster! And they for free test from Canada. So don’t worry if the number reported isn’t that low. There is indeed a very interesting note that must be considered: if the target server is slow, then that is reported as “Server is Slow” under Additional tips in the header of the report generated. If that “note” does not disappear after the optimization process, well changing hosting it is now worth. And if your server is slow and maybe it is a shared hosting account or a shared VPS, do not give attention to Page Load Times when there is high traffic, during week-end days or in the evening, regarding server location. In those moments you’ll get what the poor server could give you, bad results.

WEB PAGE PERFORMANCE TEST

is a very detailed tool and gives the possibility to choose from a great variety of test location servers. Full Grade A is the target, score 100/100 is possible on all metrix, CDN excluded, if it is not used.

ABOUT PAGE SIZE & NUMBER OF REQUEST.

The optimization process has consequences on the overall Page Size and on the number of Requests (and this parameter is essential until your server is HTTP/1. If you are already on an HTTP/2 environment you’d probably won’t be here). Both these parameters affect speed, the less those ones, the lower the Load time, of course. But the first one, Page Speed, affects the financial aspect of your website also. As soon as your website speed improves, all your rankings will improve. No wonder to see more traffic. And serving lighter pages will save you bandwidth, and bandwidth costs. Either if you buy it from your hosting provider, or if you pay for a CDN, it has its impact on financials. So speed optimization is website financial savings.

NOTES FOR SLOW SERVERS OR LOW RESOURCES, DUE TO HEAVY LOADS.

If the server is slow, or is performing low resources, due to heavy loads, caused by traffic near server limits, well the test results can be altered. This can be easily showed on the server we used for this test, on European weekend evenings it stalls, it’s very poor.

NOTE

   Obvious.

But not so that much…

PAGESPEED MATTERS, ABOVE-THE-FOLD

With WordPress it is really hard to serve little pages, with little HTML code, little CSS and little JS. This is how WordPress is built. So let’s do as Google do: let’s give our main interest to the optimization of the above-the-fold content, the first served to the visitor. That must be fast, blazing, with no flashes (WebPagetest tests will show exactly this, providing a rich diagnostic information including resource loading waterfall charts, Page Speed optimization checks and suggestions for improvements). That’s why our works are very efficient. If the rest of the page takes seconds, it’s not a problem. If you are serving a landing page type, with lots of content, of any kind, full of social interactions, Facebook timeline, Twitter, G+, widgets, animations, videos, for instance, well it won’t have a low entire Page Speed time. The visitor is engaged by the speed of what lies in the above-the-fold content, and will happily continue to read the rest of the page. One easy trick done by the experts, is to forge the pages with a very large and lite image, little logo, essential menu… just to render blazing fast. Maybe all the opposite of megamenu, image sliders, timeline sliders… that usually are the main components of the most appealing WP themes.

INTERESTED IN OUR SERVICES?

We will get in touch to help you satisfy your will.

Please send us your WordPress site url, and we’ll make a free analysis of your site.

If you want, you can follow us on Twitter Twitter_bird_logo_36x36 - (we) Optimize Your WordPress Site - Optimized Image to direct message with us.
You maybe scale up the waiting queue and have your job done sooner.
https://twitter.com/optimizeyourwp

REAL LIVE OPTIMIZATIONS

Some live optimization jobs, some tests, screenshots, numbers. Maybe the most interesting part…
WordPress Test Server Environment, optimization configuration comparison

WordPress Test Server Environment, optimization configuration comparison

Report of the test results conducted on the low budget shared hosting services. From not optimized to full optimized, maximum speed and target achieved.  #all¯
WordPress Test Server Environment, other optimization configuration jobs

WordPress Test Server Environment, other optimization configuration jobs

Another report for other jobs results conducted on purpose on low budget shared hosting services. WordPress faster as server can be.  #all¯

PS: Please note that the metrix scores results are not a fixed number. They are elaborated by the various sites as new each time a brand new request is made, so they can vary. Especially the bad results, they can float a little. Once a site have been optimized its high values do not change. So you cannot attend to have an exact reproducibility of the numbers we have published in this report.

Google, PageSpeed, PageSpeed Insights, GTmetrix, YSlow, Pingdom, WebPagetest, and others are registered trademarks of the respective owners.