The festival of WordPress
January 22, 2021

This is an archive of the January 2021 event

5 Most Popular Recommendations to Speed Up Your Website that Don’t Work

If you ever googled anything related to “how to speed up a website” you probably noticed that there are few recommendations that are repeated almost in every article over and over again.

I’ve picked 5 of the most popular tips to speed up your website that don’t work. How do I know? I’ve been speeding up websites full time for the last 2 years and was able to observe the impact of different efforts on a daily basis.

I will show you:

  • why exactly these 5 popular recommendations don’t work
  • how you can check out if something makes your website load faster or slows it down (be careful, you might want to stop paying for some services you’re using currently)
  • actionable steps you can take (whether you’re a site owner or a developer) to speed up a WordPress site

After this talk, you will never waste your time again on things that don’t work.

Speaker: Sabrina Zeidan

Time: 12:00pm UTC
Region: EMEA
Stage: GoDaddy Pro Stage

Hi, my name is Sabrina. I’m from Kiev Ukraine. I swim, I like spaghetti. I drink lots of coffee, but also I speed up WordPress websites. I do this every day. I’ve been doing this for the last three years. And during these years I never ever had a client who hired me before they tried something to speed up their website by themselves.

It always sounds the same. Hey Sabrina a while ago, we noted the towel website is slow. We tried this and that and that thing. And another thing and another thing after another thing, but basically it stays the same. What are we doing wrong? I made a list. So I made a list of the most common recommendation that you will find if you Google something like how to speed up my website, the recommendations that don’t work in.

Most of the cases I’d like to share those with you today. And I will show. Exactly why those recommendations don’t drop. Hopefully after this talk, you will be spending no more time, money, and effort on things that don’t work. But first let’s take a quick look. What happened? What is happening when the page is loading?

Oh, did you enter herbs that address or more likely you click the link on another website? Your browser makes a request to the server where the website is hosted, asking to send the data. Then server processes, the request, execute speech. The code of the website gets data from database and have the page ready for you as HTML five file, which includes text.

Font files to display that texts, CSS files to make it look nice images, JavaScript files to empower all the functionality except of these basic things. In order to serve content, to visit them. This patient needs to connect with other websites and services for purposes like animation, sliders, chatbots, messaging, apps, payment, gateways, newsletter, subscription , uh, videos, social sharing, Emily’s IX and other analytics.

Ads. So I don’t see two comments, one, one that would exist him. For example, your website would connect to MailChimp server introduce itself amongst you would response. Like, yeah, I know you, I recognize your API key. I will send you a proper data to display subscription form on your website the way you wanted to.

And yeah, I will add your new subscribers into your account. It’s not random ones. Uh, and so some images since it’s CSS files, JavaScript files would be. Requested and lauded from MailChimp server. No, I’m the same way. Uh, with other third party services, the more functionality like this you have on your website, the more requests you need to make, the more data you need to lot.

So what are this functionality? And therefore this requests are stores at everything that happens on your website comes from WordPress core theme and plugins. Considering there is a whole bunch of things we need to request process and display. It sounds like a reasonable idea that to make our website faster, we need to reduce the number of those.

Right. We’ll get to a top five recommendations that don’t work in a minute, but first let’s take a look at this two websites. Here’s the first one, it’s a popular and use portal with lots of ads, popups, sliders, videos, tones of analytics and Atlas least appalled trend in use. Um, this all causes into 515 requests and five megabytes of data.

And Kira is the second one. It’s a tiny  website or even distance card of someone who works dogs in the small town North from London. It has one large image and three smaller ones. If you look so text a couple of reviews and that’s it. No, pop-ups no ads, et cetera. And understandably it has just 17 requests and rates less than 1.5 megabytes.

Let’s get back to our recommendations. We’ll start with number five, four and three altogether PR prevent proven for press from making additional requests, disabled, unused plugins, combined requests, as they all are aimed at the same thing to reduce the number of requests. And page size. So the idea is the following.

If we make listen requests and our page waits less, we need less time to process those requests and then let’s tend to load everything. So you follow this advice, you will go home to prevent WordPress core from making additional requests. Um, you , uh, post a few lines of code into, into your functions, PHP file, or use plugin for that.

And then the next recommendation to disabled , uh, unused plugins. So you go to plugins page. You. Look for plugins that are activated, but not actually in use, you deactivate them. Um, then , um, to combine CSS and JavaScript files, almost every caching plugin has a feature for that. So you just go to the kitchen bracket that you use and take the books there to combine CSS and JavaScript files into two separate files.

But. Them then you check the number of requests and it has gone down. But as soon as you check the,

but as soon as you check the timing, you’ll see that the actual time that use it has to rate before interacting with your website. Um, didn’t really change. It’s still the same, which means your website hasn’t become any faster conversion rate hasn’t been improved and use experience hasn’t become any better.

Why it didn’t work first prevents WordPress core from, from making additional requests. If we’re, if WordPress core was responsible , uh, for your website being slow. All the websites that use word press would be slow as well. Cause we all are using the same quote, right? But that’s not what’s happening. I made a fresh install of the latest version of WordPress.

Um, this is how it loads in mobile device. Uh, 3g internet connection. It  needs 2.9 seconds to get fully loaded on iPhone eight where the three G internet connection in Dallas. It’s makes nine requests and page rate is only 42 kilobytes. Let’s just constant, full paint on mobile by Google page speed insights is 2.0 seconds.

And this all without caching without any optimization techniques at all share at horsing and free SSL certificate, freshly installed. WordPress is fast. Those requests. And page rate don’t come from WordPress core. They come from themes and plugins that we use to get the functionality we need. So if there are so many CSS and JavaScript files needed for all that functionality, what we can do about this and other popular recommendation that you probably saw a lot is to  combine those requests into one.

I mean, into two. So say you have , um, 10 CSS files on your page and turn JavaScript files on your page. And do you combine them into two separate files? One, one for CSS, another one for Chelsea group. So instead of making 10 requests, you’re making just two requests. Oh , well, it was a useful recommendation years ago.

Well, most of servers we’re using and she’s be 1.1 protocol, but you would have hot times finding Holsten. There’s still uses. It should be one protocol. As most of them use HTP two nowadays and here comes the big difference. It should be one would process 10 requests one after another. So it absolutely makes sense to minimize the number of them.

While it should be two is fully multiplexed, which means it allows multiple pilots and request to be transferred at the same time. And combining files will have  less of an impact on loading times. In many cases, it would even make it worse. Recommendation, number three, disabled unused plugins. This seems like a reasonable one, right?

So you go to a plugin space and she allowed which plugins are activated and not being in use actually. So you can sample them. But the thing is, it doesn’t really help lets me explain at the time when I was working one of the clients sent us a support request to help them to speed up their website.

Then side had 400 plus plugins activated, including focusing plugins at the same time. It was an absolute champion. I don’t know if anyone has broke this record less wrong, but that was the most impressive one that I’ve seen. If you’re listening to this talk at the moment, chances are that your website or your client’s website is nothing like this.

If you’re listening to this talk, it means that you are interested in keeping it nice, clean and efficient. I would assume that you keep the core themes and plugins. Updated you care about what you install. You read reviews before activation themes and to avoid bloating the website at all costs. So if you go through the list of the plugins that are activated, but not in use, it turned to be a slight decline.

It won’t be lectured plugin or popup plugin, or contact or contact form and so on because if you have them activated, they are there for a good reason. Your findings would probably be something like manual backups. So should place import export plugins. In other words, though, those that are used occasionally and can be disabled before flawlessly.

But the thing is that those plugins don’t do much either unless you interrupt with them. So once you disabled them, not much has changed in the site’s performance. So yeah, if your website is a Frankenstein like these. The idea of disabling unused plugins might be a useful one. And then you are, but perhaps it’s not your case.

So what is your case? Them chances out your numbers. Uh who’s to these speed index, which indicates how quickly the page displays content or user can interact with around six seconds request count. 111 and page rate , uh, close to three megabytes. These are average numbers for the K reported by Google.

Remember those two websites we looked at before, let’s be remind you. The numbers use website has 515 requests and four megabyte and speech rate is four megabytes and the Walker’s website has  17 requests and it rates 1.3 megabytes. So let’s take a loop. How would they both, although gene from the same test and server, same device and same internet connection.

This is how use website with 500 requests is Lord. Yes and no humorous it tiny new website with 17 requests.

User has to stare at a blank screen before anything starts happening for quite a while. Lots of content full paint on the news website is 2.6 seconds while alleges content full paint and dog walkers , uh, site is 7.8 seconds. What is content full paint is the , uh, Main content that you see when, when you just load the  website on the first viewport.

So despite of being quite lightweight, this super simple, super light website makes the users two ways almost for eight seconds before they can see at least anything on the screen. As you can see it needed as you can see it’s near the size of the page. No. The number of requests that makes a web page fast or slow.

Let’s see how those requests are executed and how that data is delivered. You probably know what is this it’s called waterfall. It illustrates everything that is happening on the page. You won’t be able to see details here, but you can see the general picture. The first screenshot is a tiny websites.

Waterfall. And we have all those 17 requests displayed here. Look at this large blank area, nothing is happening for a long time. And then some requests  get processed. It takes 11 seconds before users can interact with this lightweight. Super simple website sevens green shirt is just a part of use books.

That login process. It has 500 requests, right? But only 12 of them get executed before the content is displayed to use them. So they can interact with that. And this happens in just two seconds. So while the user is already enjoying the website’s content, the rest of the content is loaded in the background.

Nice. Isn’t it. The next recommendation we see a lot. Number two in our list enabled CDN first, how content delivery network works. Say your web host has a server in longterm without TDM. When a user requests your website, it gets a delivered from London. We sit down when user requests, your website, it gets delivered for one of the sedan servers across the globe.

No. Which way is fostered? Let’s see it highly depends on how fast your web host servers are, how far stadiums servers on where Sudan servers allocated and how sedan decides which server to use to serve your website when the request comes in. But even these are not that. Those questions that’s , uh, mostly important.

The most important question is where your audience is. Take a look at this website, it’s sort startup reply. I O it’s a sales engagement platform that automates personal email outreach calls and tasks. If you check the location of their users, you will see that the majority is coming from the us with another big chunk of users coming from the UK.

As you can imagine both locations are important for their business and they want their website to be loads and fast on both sides of the product. It makes sense for them to use sedan to serve it equally fast everywhere. Now, if you run the same chair on from common cause website , Uh, pizza chain with Jewish Russ pizza, good vibe and incredible Ivanka to visitors do find out that 93% of their visitors come from the UK.

And there is no one to actually it’s UK chain, right? So they don’t care how fast their website is, is in Australia. They want their website to be as fast as possible for visitors from the UK. They would benefit the most if they won’t be using sedan. But instead, if they make sure that they have a FoST server located in the UK, the closer to London, the better as more than half of their restaurants are located in London.

If website audience is not spread all over the world, but comes from the certain area, chances are that sedan won’t make use experience better. Most likely you would make it worse. I wrote a detailed article about eat for roadways book, feel free to check it out. And the last one , uh, but not the least. And my personal favorite one , uh, change your repost or upgrade your question.

Glenn, if you, years ago , uh, one development company contacted me , um, They just finished development of a brand new website for their client. It looked beautiful. It functioned as it was expected to, but it turned out to be tremendously slow. They just couldn’t have the, to climb that way. So they implemented some caching.

They did some basic performance optimization things. Um, it didn’t work. They contacted their hosting provider a good one by the way, and ask them what can be done about that. Um, they would recommend it to upgrade their hosting plan and they did that and eat. Really mitigated issue a little bit, but just in two weeks it became a slow, as it was front-end was still slow.

Backend was hardly usable because if more than two people would log in at the same time, just nothing worked. So they hired me and told me the whole story. And I was really surprised at that time that they are hosting to provide a support, would recommend them to upgrade their fun. Instead of doing just that one thing that would really help them to fix the problem, to profile their code, to find the issue, fix it and leave happily.

But that’s in an ideal world. Right. Um, as I learned later, it’s like common response. If you ask you web host, What can be done about your site speed? They would recommend you to upgrade your hosting plan, though. It doesn’t seem. Nice from this side, I’m convinced that it’s our responsibility as developers not to be fooled like that.

If you’re newly created website that doesn’t have any traffic yet, or is hosted on staging Lord slowly, clearly the issue is with your code, not with resources on your server. Great. In hosting plan or move into another host with more resource, for example. My to mitigate the issues you have in your code, but it would never fix them.

So you’ll just find out quite soon. It’s slow again. I quick note, obviously there are some holes that are much slower than others, but if your site’s code is fine, you won’t be experiencing any issues until you get some decent traffic on your website. So this world, top five recommendations helped to speed up your website.

That don’t work first preventable, press call for making additional requests. You saw that , uh, you freshly installed WordPress it’s super fast. It’s super lightweight. Um, even without any caching and optimization techniques on share at Horsley was free as a salted 30 bigger certificate. It loads fast.

It’s us who make repress slow. When we install theme plugins, we add additional functionality. There’s obviously we need to have one on websites, but the WordPress core itself it’s super light and super fast. And other one, the fourth one does used to plugins. Yeah. Great advice. Unless you don’t have anything to disabled.

The third one combined requests. Great advice  as well was five years ago , uh, with , um, servers that we’re using the 1.1 protocol, but now you probably , uh, your website just drove the horses on the server. That’s your, that uses HTTP two protocol and the district mentations recommendation makes no sense at all.

Another one enabled sedans. Which might work for some , uh, websites that have glance over the roads and visitors coming from all over the world, but for local websites, or even for works that have visitors coming from the specific area, this would make their websites slower, not faster. And the last one change or upgrade well cost.

You should. Especially that you could send you upset. Oh, if it’s not you, but you don’t have like really nice amount of traffic on your website, you would post it less than you should suspect first profile your coat, make sure nothing goes wrong there. And this is not the case most of the time. And then look at , uh, your host , uh, resources , uh, resources starts.

Make sure they. Uh, consumed already. And then maybe think about upgrading your hosting plan plan or move into another host. So all those recommendations are bet let’s have a little break. Here are two guys in Tampa, Florida trying to break into a safe in taco bell restaurant. They brought some tools. They make some noises.

Um, they flush some nights, but still no luck. To break it wide open. And when that doesn’t work, they bust out a tool hoping to bust in the safe, a portable circular saw with sparks flying everywhere. So that’s, you can open that’s kind of digital safe using just a hammer. It’d be a can your hands or even potato.

Yeah. You can do that in case you understand how digital safe works and in case, you know what you are doing. And no, those recommendations are not bad. It’s just that they’re not suitable for each and every case for most cases, actually. So how do you find out what is suitable for your website? What will help your website to run faster as you probably already understood?

The only way to really see what’s going on there is to look into waterfall. It shouldn’t be a guessing game. Everything isn’t there in waterfall. Also here, if you teach that would help you to understand your website speed better. If you’re a developer, they would help you to learn. And in case you are not a developer, they will help you to hire someone to do the work for you without getting pooped.

So here are a few tips.

The first one is to use right tools for specific purposes. I use page speed insights for cook for quick checks. I use lighthouse in Palmdale students for analysis. Um, I use a web page test.org for real user experience testing because it lets you choose between , uh, different , uh, testing server locations.

It, it lets me choose different connection, connection, tabs, and devices and everything. So if you need to choose like real user experience, this is what page does that org , um, That no matter which tool you use, forget about course at all. So speed is measured in something very specific in milliseconds.

When you test site speed, always look at largest quantity for paint or other metric that this place, a user experience, for example, speed index, or tend to interrupt it. The next one test important pages, a few episodes, not homepage. I mean, Home page is obviously important, but you have all the important pages on your website.

For example, if it’s a WooCommerce store you would like to test , um, single products page, you want to test your landing page. You want to test your Checkout page and others, not if you’re testing homepage only, you can see the entire picture of your backside test. All important pages on your website. And the last one, keep an eye on it.

Your website does not stale. It’s alive. You change something. Uh, your hosting provider changed something. You installed some, you plug it. Uh, some of the plugins that you had before , uh, was updated, it changes all the time. So do you need two more inter yes. Your site speed. Uh, Constantly to see the , uh, changes here is the tool that I use for to medic every day.

Monitoring it’s called SpeedGuard. It’s a WordPress plugin. You can have as many pages of your website. More, you start every day as we need. Those can be pages of different types, posts, pages, WooCommerce, products, events, any other custom post types, and also. Terms taxonomies and category or archives. It uses page speed, insights, API Torontos, and it measures largest content full paint.

It calculates the average across your entire website and identifies you in case it needs your attention. Um, that’s all for free. I was looking for clogging like this and I couldn’t find one. So I built this one. Um, you’re welcome to try it out and please feel free to reach out and let me know if it’s made your life a little bit easier as it did for me.

Thanks for listening. You can get in touch with me via my Twitter, my website or email. My name is Sabrina Aidan. This was five popular recommendations to speed up your website. That don’t work. If you have any questions, feel free to reach out. Um, I hope to see you in a chat in a couple of minutes. Bye.

Share this session

Share on facebook
Share on twitter
Share on linkedin
Share on pinterest
Share on email