How I Improved my Website Performance by 72% Without Touching a File

speed*

A while ago I have written a post on webhosting issues that harm SEO for WebHostingBuzz. It was a list that contained an issue many businesses still ignore at their own peril: most people abandon slow loading sites. Mobile users are even less patient.

Despite website speed or performance being an official Google ranking factor for most businesses load speed is still just an afterthought.

The guys of WebHostingBuzz challenged me to write another post about speed suggesting that they are faster than the average host.

I was apprehensive at first, after all I usually don’t write about topics others want me to. As a blogger I chose my topics myself. Also here was a claim I couldn’t just support without proving it’s true. So I decided to test whether simply changing a web hosting provider can accelerate your website and enhance load speed significantly. As the headline above already says, it can. I was astounded by results.

I successfully improved my website performance by up to 72% without touching a single file.

The Set up

How exactly did I do it? How can you be sure I don’t err? Well, I used a free public tool to measure my success and you can test it yourself. It’s the Pingdom website speed tool I have recommended repeatedly in the past.

To be able to compare my current hosting with the new one I set up a mirror of my site.

I cloned the English version of it from onreact.com/en and uploaded it to WebHostingBuzz servers on onreact.net/en – then I made sure that everything is exactly the same with the exception of the hosting of course. I had to remove my Piwik code because it was installed locally on onreact.com

I copied the images, the scripts and stylesheet over to onreact.net while third party tools for analytics and webfonts remained where they are on their own servers.

So you may argue that the test is not completely objective as these separate tools can affect the load time as well, That’s true to some extent but on the other hand it’s more realistic to measure that way. After all every site uses some third party tools like Google Analytics for instance.

So I didn’t optimize anything. I just removed to Piwik in order to be able to compare the two sites on equal terms.

 

The Tests

pingdom-speed-test

At first I couldn’t believe what happened. I selected the same location for both sites: New York. I hit “test” at roughly the same time during the day (afternoon here in Central Europe but morning in NYC). So I kept hitting test on different days and repeating several times in a row. The load time would vary slightly but not really significantly.

I wanted to make sure that there are no very outlier results but couldn’t find any. Of course I also checked from the two other locations available. Dallas showed an even bigger difference. I also looked specifically into the items that got loaded.

Many in the SEO industry suspect that for Google the most important ranking signal is the actual server response time and not the whole load speed of the site.

I don’t think it’s either or, I guess it’s both important. Anything else wouldn’t be very logical. What exactly loaded slower or faster? Onsite resources like images or rather external items like analytics or webfonts?

My biggest image is just 55.75 Kilobyte but still the differences in load speed were enormous. See below!

 

The Results

In most cases the onreact.net mirror has been below 500 milliseconds. That’s fast as hell. Pingdom itself says that it’s faster than 98% of websites. This is a typical result: 428 ms.

onreact.net-speed-428onreact.com-speed-1.55

The original site at onreact.com was usually three times slower, in most cases above 1.5 seconds.This is a typical screen shot where it takes 1.55 s to load the same files as above:

Most surprisingly it’s not only about the server response time or the third party tools.It’s really about onsite load speed the web hoster is directly responsible for.

The above mentioned 57.5 kb image loaded more than 12 times faster on the onreact.net mirror!

onreact.net mirror

onreact.net-image-load-55

vs onreact.com original

onreact.com-image-load-697

image load time.

 

Why is the original site so slow?

Now you might ask yourself: why the heck does he use such a slow web hoster in the first place? Well, it isn’t that slow. It’s allegedly one of the best hosting companies here in Germany.

When testing from Amsterdam in the Netherlands which is nearby the German based onreact.com tends to be faster than the onreact.net from the US server. Still the difference is not so big. Also the results vary quite a lot each time. Sometimes the original is faster by 50% than the mirror sometimes onreact.com as fast as onreact.net.

WebHostingBuzz has also British servers in the UK I could use for a better comparison when it comes to European load speeds. Next time maybe. Looking up a Germany based site from America is apparently much slower than from a hoster based in the US. So at the end of the day it’s mostly the geographic location and only to a smaller part the fast hosting from WebHostingBuzz that makes the difference.

For me personally the best choice would be to move to WebHostingBuzz.

I consider it as my current host is pretty annoying lately anyway. My blog audience for example is mostly in the US. So hosting in Germany out of laziness is not a good idea it seems. When you serve an international audience move where the audience resides, here it’s the USA.