Running tests on popular online speed testers like WebPage Test, GTMetrix, Pingdom, and Google PageSpeed Insights can give you radically different results.
Cut through all that noise and learn the basics of running an accurate speed test and see what’s dragging down your page load time.
First, let’s have a quick overview of the different online testers and why I rely so heavily on some and think others are a waste of time.
This is my workhorse simply because it yields the most accurate results, and more in-depth data to get to the bottom of what’s dragging your site load speed down.
It is the only free online tester that allows you to run multiple concurrent tests, and that’s radically important to get a statistical average.
With the rest of the testers, you have to run a bunch of individual tests and manually average all the results.
Plus, no other online tester gives you as many location and device choices, including multiple mobile devices and connection speeds.
And, it is the only tester that accurately runs HTTPS sites on HTTP/2, which shows a more accurate picture of your site elements being delivered in parallel.
This is the tester I’ll show you the basics of running at the end of this post.
This tester is a super companion to WebPage Test.
It gives you results for the most important elements tested by both Google’s PageSpeed Insights and Yahoo’s YSlow.
Actually, it does a better job of it than either of those two testers do directly.
You’ll need to get a free account so you can log in and test from a location other than Vancouver, Canada.
While GTMetrix does give you a mobile speed test option, the only location you can run it through an actual Android device is Vancouver.
Plus, it’s a one-off tester, meaning that you have to run multiple tests and manually average the results.
FYI, most money-making sites will never get an A score on YSlow. I did on one of my test sites, just to see what it took and it’s insane!
This tester has the least number of test locations, and they are often backlogged.
Plus, the on_load point for testing above the fold rendering is far shorter than any other tester.
This is why it seems your site loads faster on this tester and will make you think you don’t need improvements.
I don’t even run this tester except to show site audit clients how much it misses in delivering the kind of data we need to improve speed.
It is also a one-off tester and you will likely get wildly different results on each test.
Google PageSpeed Insights
Forget about it!!!
This is, by far, the least accurate and least helpful tester on the planet.
In 2016 we all lost 15-20 points because we weren’t using AMP.
In 2017 we all lost 15-20 points because we weren’t using SVG images.
In early 2018 we all got 20 points back on mobile, and nobody knows why.
The one and only thing you want to pay attention to on this tester is the render of the mobile version of your site. To my knowledge, it is the one Google is using, and not the Mobile Friendly tester’s version. (I know, insane, right? But they do check the site differently.)
READ: Stop Chasing Scores on Google PageSpeed Insights for more on why not to rely on this tester.
Chrome Dev Tools
This is a super tester you run directly from the Chrome browser.
It gives a pretty accurate test of how fast a site runs at your location with your internet speed connection.
It also has a mobile tester, but is a best-guess on that speed, as it is simulating the connection.
Still, it’s a super handy tool I use often for header response checks and to see which caching and security headers are in play on a site. Plus, it’s fantastic for locating mixed media issues on an HTTPS site.
What the Testers Don’t Show
Most of what I check during a site audit is at the host level.
Security and performance go hand in hand.
Plus, a clean environment is critical to site speed.
So, what causes all the messy files, folders, and database tables?
Here’s an example.
When you delete a plugin, it comes out of the plugin list you see on your site.
But, it may leave behind orphaned files, folders, and database tables.
Plus, it could leave behind conflicting directives with the new plugins you are using.
That’s especially true of caching plugins. They spew their directives, files, and folders everywhere.
The same thing happens when you move to a different hosting environment.
All of that junk needs to be cleaned out.
And your site needs to be properly secured from the root up.
Security will also affect your site speed.
Once you eliminate the bots from chewing up all of your hosting resources, you’ll have more left over for your human visitors.
That security will also save you on hosting costs.
So, running the online testers are a good thing to do, but only one part of the picture with regard to what’s slowing down your site.
Let’s Run a Performance Test!
As I mentioned, WebPage Test gives you a TON of data and is my go-to tester.
Learning to interpret that wealth of data is well beyond the scope of this post.
Here we’ll cover how to run the test accurately and have a look at the most important data.
If you want to go deep into site performance testing, see Level 4 of my Webmaster Training tutorials.
Step 1. Chose a typical blog post to test, not your home page.
I’ve found that blog posts will show far more speed issues and help you eliminate more drags.
Step 2. Go to https://webpagetest.org
Step 3. Choose URL, location, and browser
Input your blog post URL.
Select a location. If you’re audience is mainly in the U.S. then select a location in the middle.
I like Lincoln, Nebraska, or Dallas, Tx.
Step 4. Advanced settings
Click the down arrow on Advanced Settings to expand that section.
Leave the first field set to a Cable connection.
Select 3 concurrent tests. This is the minimum you need to run for a good average.
Select First and Repeat Views so you can see how well your caching is working.
Turn off Capture Video.
Leave Keep Test Private selected.
Step 5. Start the test
Click the button in the upper right to start the test.
Sit back and wait. This is going to take a few minutes.
Interpreting the Test Results
Forget about them!
It’s not about the scores, it’s about what goes into making them up.
Most of them are based on percentages.
Not everything that goes into them is under your control either, like anything that comes from the outside world.
For instance, the Google Fonts your theme uses are coming from their CDN. If it is having a slow day, so is your site. And that will be affected in the score.
Same for Google Analytics, Facebook Pixel Tracking scripts, and Gravatars.
In fact, I’ve optimized sites to the point that only the things from the outside world were slowing things down. And since the percentage went way up on those outside items, the score went down. But, the site was immensely faster.
THIS is what you really want to pay attention to!!!
Let’s break it down.
First Byte Time
This is also referred to as TTFB, or Time To First Byte.
You want this to be .5 second or less.
Long TTFB could be an indication of a poor host. But, I’ve seen LOTS of site elements affect it. So, don’t jump hosts before you take proper measures to speed up your site.
You want this to be under a second, if possible. Some outside influences will affect it, though, like Google Fonts, so you may not be able to easily change this. But if it is over 1.5 seconds, you definitely want to have a look at site elements you can control and fix them.
This is to the on_load point. On WebPage Test, it will be longer than any other tester, especially when you compare it to something like Pingdom. For desktop testing, you want this to come in under 3 seconds for sure, but aim for under 2 seconds if you can.
Fully Loaded Time
This is the time it takes to load every element on your blog post. You can actually get away with a pretty long load time of 7 seconds or more if, and only if, you do what is necessary to fix render-blocking elements above the on_load point. That way your site’s perceived speed will be much higher too.
This is HUGE. It represents how many elements it takes to load your full post.
The average is 140.
I’ve seen sites that have 800 or more requests.
There is no speed tweak in the world that will overcome that many requests.
It is imperative that you do what is needed to lower it.
That could be as simple as swapping out a few plugins.
See this post on how the Pinterest Widget (or script) overloads your site with requests.
And see this post for widget alternatives that won’t slow down your site.
This is your actual page weight. You want to shoot for 500KB.
If your site is 1-2MB, you HAVE to take the weight out of it if you want real speed.
You are NOT going to push a mega ton boulder up a mountain very fast.
You WILL push a pebble up quickly.
All of the numbers in the Repeat View row should fall dramatically (down to 1/3 or less) compared to the First View row.
This is an indication of how well your caching is working.
If the numbers aren’t radically lower, you need to do 2 things:
- Discover what types of elements are loading anew (they will likely be JS files)
- Check your cache settings everywhere (at the host, local plugin, CDN like Cloudflare)
Reading the Waterfall
Remember when I said that the scores don’t matter, it’s what goes into them that counts?
Well, the Waterfall is where you see exactly what goes into those scores.
It’s also where you see every single element, or Request, that is loading on your site, and the order of that load.
That order is a huge deal for perceived speed and at the heart of what it’s going to take to make your page load faster.
2 Places to See the Waterfall
If you want to see the Waterfall for the median run summary results:
Click the Run link in the First View section.
It will drop you down to that run. Then click the Waterfall image.
Or, just click the Details link in the top menu and it will take you to the Run 1 Waterfall.
There are 3 elements that are going to contribute the most to a slow page load time.
Those critical elements are:
A whole bunch of CSS and JS is a good indication that you have a bloated plugins, and perhaps a bloated theme.
Your best bet is to track those down and reduce the weight by either removing them or substituting for a lighter plugin or theme choices.
Image reduction and optimization is the #1 way to speed up your site.
(FYI, it will show you how to run this same WebPage Test, but how to make it run faster, which is less accurate for full page load tests. And, it shows you a secret link to focus directly on your images.)
Length of Bars
The length of the bars is to scale.
Be sure you pay attention to the number at the end of the bar, not its length to see if the actual speed of that element changed.
For instance, the first bar is mostly your host compiling the HTML document of your page.
So, if your site is loading in 20 seconds and you speed it up to 10 seconds, that bar length might suddenly be twice as long, yet it took the same time to compile.
Order of Elements
This is HUGE!!!!!!!!
The order elements load makes a world of difference to your site’s perceived load speed.
If you have a ton of render-blocking elements like CSS and JS above your images, then visitors are just staring at a wheel turning on a blank screen.
If they have to wait too long, they’re just going to leave before they see anything.
THIS is why you MUST reduce all that weight and reorder the elements so visitors see what they came for quicker!!
Click to See the Element
The left column of the Waterfall truncates the path to the element.
But, there are easy tricks to see what each element is.
First, hover over the link.
The full path will be revealed.
Or, just click on it and a box will pop up with all the info.
You can also click on the Object tab and see the element, if it is something that can be rendered, like an image.
Get More Info
WebPage Test has lots of menu links at the top with deeper info on your page elements.
Performance Review will show you how well your caching is working.
Content Breakdown will show you the element bloat and which ones you need to reduce the most.
Domains will show you how many elements you’re pulling from the outside world so you can reduce their drag.
The data in WebPage Test is deep. And it may make your eyes glaze over.
But, it’s only one piece of your site’s performance puzzle.
You don’t have to figure all of this out alone.
You’ll get a 20-30 page report and a 2 hour, non-geek-speak chat to go over it.
You’ll clearly see everything that’s causing your site to be slow.
And then you can make informed decisions about how to speed it up instead of throwing darts in the dark at it.
Want to Learn All About Site Speed?
See Level 4 of my Webmaster Training courses.
These tutorials are made specifically for designers who need to learn more about site tech.
It’s training that pays for itself too!! You’ll be able to finally land those in-demand jobs and offer more services to your clients that will separate you from the pack.