The Speed Study Goal
My goal with conducting these Speed Studies is to find out which strategies are the best at improving our site speed.
I’ll implement each strategy using the following approach:
- Setup the strategy on all 8 of our live test sites
- Then for the following 3 weeks, gather the load times from our Desktop Speed Test using GTmetrix. Each week we will test from a different location, starting in Dallas, then London and finishing off in Sydney.
- Next measure the impact of each strategy by comparing the average speed before and after
- Ultimately decide whether the strategy works?
EWWW Speed Study
With our 2017 baseline now setup and an Imagify study just completed, lets apply our second speed strategy of optimizing the images on each site. What is expected is the page size will reduce, what we are really interested in is how much speed improvement is possible. The plugin we will use for this Speed Study is EWWW Image Optimizer, which is a freemium offering with over 400,000+ active installs.
To kick off this EWWW Speed Study, the first step is to remove Imagify and then install the EWWW plugin. Next I will follow these installation & setup steps for each of our test sites.
I have now followed this process on all 8 of our Live Test Sites.
Next up is 3 weeks of testing – I am curious to see how much our speed differs between Imagify & EWWW?
The Results – 13 Mar 2017
After 4 weeks of testing, lets take a look at how much of an impact installing EWWW has had on our 8 hosts.
So how did EWWW perform? The average load time across 3 different testing locations was 3.38s, which is 13% slower than Imagify.
Let’s also remember that the speed shown during the Imagify Speed Study was the same as when we had No Optimizations setup. With that in mind, when we see that EWWW is actually slower than Imagify, we have some problems!!
Even when we dig into the 3 locations we test from, all 3 experienced slower results. On top of that, there is also no difference in the other factors such as Page Size, Google & Yahoo Scores.
When it comes to making a call on whether EWWW is helpful, the answer seems as though it is a clear NO. However, in the section below where I break out each of our hosts, this overall conclusion will need to be held back – read on to find out what I’m going on about…
Individual Host Performance
Let’s now take a look at how our individual hosts performed:
- The results are mixed when it comes to our individual hosts:
- Bluehost, GoDaddy & InMotion Hosting experienced minor improvements of between 1% & 8%.
- SiteGround, HostGator, Hub & Namecheap had minor negative impacts, ranging from 1% to 4%.
- But the big news story here, is that DreamHost fell off a cliff!! Dropping from 2.92s to 6.04s is a crazy change, and at this stage it’s hard to tell if this is related to EWWW. The real test will come over the next 3 weeks, with EWWW being changed out for another image optimizer. If the speed suddenly returns, then we will hold EWWW at fault for having some sort of conflict. If not, I will return to these results and exclude DreamHost and see where things stand.
- Update: Following the Speed Study on Kraken, it has been proved that DreamHost was at fault and not EWWW. With that in mind, I have removed DreamHost from the results table below in order to compare our Image Optimizers fairly.
With both Imagify and EWWW now put through the wringer, lets find out how another Image Optimizer will perform?
So Speed Study #11 will focus on what speed improvement is possible with Kraken?