When it comes to choosing the best WordPress Hosting, I’m bound and determined to find out which company is providing the most value. There are three major components that make up a great host in my opinion, and those are 1) Performance, 2) Knowledge, Speed, and Reliability of Support, and 3) Pricing and the overall product offering.
For this initial run, I compared several shared WordPress hosting companies. I plan to continue to add to this list and update it in 2014 as well so that people have a goto resource for choosing the best WordPress hosting company. Here are the hosts I’ve tested in no particular order:
| NAME | ACCOUNT TYPE | COST PER MONTH | |
|---|---|---|---|
 | SiteGround | Startup | $3.95 |
 | Bluehost | Standard Shared | $4.95 |
 | Inmotion Hosting | Power Plan | $4.89 |
 | A2 Hosting | Premier | $3.99 |
 | Site5 | HostPro | $8.95 |
 | MediaTemple | Grid Server | $20.00 |
 | Dreamhost | Standard Shared | $8.95 |
 | Eleven2 | S-200 | $8.00 |
 | GreenGeeks | Standard | $3.96 |
 | Arvixe Hosting | Personal Class | $4.00 |
 | HostGator* | Hatchling | $7.16 |
 | GoDaddy | Deluxe | $8.99 |
Today I want to take a look at how many of the top WordPress hosting companies measure up from purely a performance standpoint.
Now I’ll be the first to admit that having the fastest servers doesn’t make a host the best WordPress hosting company, but it certainly doesn’t hurt. If you’re looking for the best WordPress host for your website, the data here’s a great place to start your research.
The Performance Testing Experiment
Testing speed and performance of servers in remote locations introduces some fun and interesting challenges. When I first set out to run these tests, I thought I’d run benchmarks for an hour or two and then be done and would be totally ready to crown the fastest performing WordPress host.
That wasn’t the case at all.
Eliminating as many variables as possible from the test ended up being a lot more work than I anticipated, but the end result is some seriously concrete data.
To run the tests I used a load testing service called Load Impact. Load Impact fires up an Amazon server that you choose, and begins to send traffic to the site, increasing slowly over a certain amount of time.
For this particular test I sent 50 users to a testing site, increasing from one user to 50 concurrent users, over the span of 10 minutes. 50 concurrent users is a nice baseline test for a shared host. Once you hit that benchmark on a fairly consistent basis, it’s probably time to start exploring VPS and Managed WordPress hosting options. Traffic increases steadily throughout the timeframe until 50 users are visiting the site all at the same time.
Here are some of the precautions I took to keep the tests as fair as possible, even though each of these sites are spread out all over the United States:
- Each WordPress site is an identical install of WordPress 3.6 with the TwentyThirteen theme installed and a number of posts and pages (same number of posts and pages on every site). For example, here’s the site I used to test Dreamhost: http://dh.wpsc.me. All other testing sites were exact clones of that site.
- I used the same domain name for every testing site with a different subdomain. I didn’t want the chance of any latency showing up in results because each testing site had a separate domain with a potentially separate registrar.
- For all hosting accounts located on servers on the West Coast, I used a testing server located in Ashburn, VA. And for all hosting accounts living on servers on the East Coast, I used a testing server located in Portland, Oregon.
- I did everything possible to make the physical distance traveled the same across all tests.
- All caching and plugins were disabled on each site
- Each test was run 5 times with the best result of each posted here.
Overall I ended up with a really nice set of data that gives some very good insight into which hosts make performance a priority, and which hosts have some work to do.
So Which Is the Best WordPress Hosting Company?
Without further ado, here’s what I found from each host. The green line represents the increase in traffic, and the blue line represents the response time of the site as traffic increases.
SiteGround
SiteGround was one of our top performers, and is especially impressive considering the $3.95 price point for their StartUp hosting tier. It does have a limitation of one website at that price point, but considering the average response time was ~700ms all the way to 50 concurrent users with no real hiccups, SiteGround offers a great value.
Testing Server Location: Dallas, TX
SiteGround Server Location: Chicago, IL
Max Response Time: 1.79 seconds
Minimum Response Time: 669.9 milliseconds
SiteGround Server Location: Chicago, IL
Max Response Time: 1.79 seconds
Minimum Response Time: 669.9 milliseconds
Bluehost
Bluehost’s performance was lackluster. As traffic increased so did response time, almost following the same steep climb. Even at lower user counts the response time jumped around quite a bit, ranging anywhere from 1 to 3.5 seconds with only 10 active users. As Bluehost approached the 20 user mark load times skyrocketed to over 10 seconds. They did come back down, but performance was still highly inconsistent with huge variances from one second to the next.
Testing Server Location: Ashburn, VA
Bluehost Server Location: Provo, UT
Max Response Time: 10.64 seconds
Minimum Response Time: 915.53 milliseconds
Bluehost Server Location: Provo, UT
Max Response Time: 10.64 seconds
Minimum Response Time: 915.53 milliseconds
Eleven2
Eleven2 is likely the smallest hosting company that we tested on this list, although I don’t have the date to confirm that. That said, performance-wise they do pretty well as a shared hosting provider. With site load times of just under a second throughout the entire test, Eleven2 isn’t a leader, but they’re definitely no slouch. The $8 per month price is only available when you pre-pay for a year.
Testing Server Location: Dallas, TX
SiteGround Server Location: Wichita, KS
Max Response Time: 2.01 seconds
Minimum Response Time: 898.61 milliseconds
SiteGround Server Location: Wichita, KS
Max Response Time: 2.01 seconds
Minimum Response Time: 898.61 milliseconds
Site5
As noted above, because of a DNS propagation issue we actually made a mistake with one of our tests so we ran Site5 through the gamut again, and again they did very well. While their minimum response time was higher than initially reported, their max response time was lower than we initially reported. Throughout the majority of the ten minute load testing, Site5’s server response time stayed steady at 750ms to 1 second with only a handful of deviations.
Testing Server Location: Portland, OR
Site5 Server Location: Atlanta, GA
Max Response Time: 1.95 seconds
Minimum Response Time: 704.41 milliseconds
Site5 Server Location: Atlanta, GA
Max Response Time: 1.95 seconds
Minimum Response Time: 704.41 milliseconds
Inmotion Hosting
Inmotion Hosting’s scores really took me by surprise. With one of the fastest minimum response times, and by far the fastest max response time, InMotion stayed right around 600ms for the entire test, which is really impressive. The graph looks to have more hills and valleys than most, but that’s because it stayed so close to the median response time for the entire test. The variance from highest response time to lowest response time is roughly 388ms, which is the best in the group.
Testing Server Location: Portland, OR
InMotion Hosting Server Location: Washington DC
Max Response Time: 836.78ms
Minimum Response Time: 478.42ms
InMotion Hosting Server Location: Washington DC
Max Response Time: 836.78ms
Minimum Response Time: 478.42ms
MediaTemple
MediaTemple Grid Server is a bit pricier at $20 per month than the other hosts featured in this post, but technically it’s still considered a shared host which is why we included it. While not boasting the fastest load times, aside from a strange hiccup at the very beginning of the test, MediaTemple was rock solid all the way to scale. Variances were 2-300 milliseconds but load times generally stayed at almost exactly 1 second, regardless of the number of users.
Testing Server Location: Ashburn, VA
Media Temple Server Location: Los Angeles, CA
Max Response Time: 4.54 seconds
Minimum Response Time: 934.07 milliseconds
Media Temple Server Location: Los Angeles, CA
Max Response Time: 4.54 seconds
Minimum Response Time: 934.07 milliseconds
A2 Hosting
A2 Hosting makes some pretty bold claims on their website, claiming 300% faster load times with WordPress. So do the results match the claim? A2 did pretty well overall, but definitely not 300% faster than the competition. Many of the hosts listed here which don’t even make claims to be WordPress hosts performed better. Their minimum load time of 455ms is definitely impressive, and it was only slightly higher than that when the test ended. Overall they had a strong showing.
Testing Server Location: Ashburn, VA
A2 Server Location: Ann Arbor, MI
Max Response Time: 1.12 seconds
Minimum Response Time: 455.17ms
A2 Server Location: Ann Arbor, MI
Max Response Time: 1.12 seconds
Minimum Response Time: 455.17ms
HostGator
HostGator’s server performance looked a lot like a pattern you’d see from a healthy EKG, until it completely flatlined. The only problem is that for web performance, we don’t want to see a line with a a lot of ups and downs, flat lines are great unless they fall off the grid completely a la GoDaddy. While HostGator returned the fastest response time of any host, it’s a little misleading because the server had essentially quit at that point and then stopped responding completely. Basically it was one last hurrah before it called it a day.
*I was happy to see that my HostGator account was housed outside of the Provo, UT datacenter where Bluehost resides so we could get a more true host to host comparison.
Testing Server Location: Portland, OR
HostGator Server Location: Charlotte, NC
Max Response Time: 10.16 seconds
Minimum Response Time: 258.07 milliseconds
HostGator Server Location: Charlotte, NC
Max Response Time: 10.16 seconds
Minimum Response Time: 258.07 milliseconds
Arvixe
Arvixe has been in the hosting business for quite a while servicing other open source communities like Joomla and Drupal, and have just started shifting their efforts to the WordPress space in the last year or so. Their results here are respectable. They aren’t blow your mind fast, but they do seem solid all the way up to the 50 concurrent user mark. They had one small spike, but it recovered very quickly and the server finished the test in heroic fashion.
Testing Server Location: Palo Alto, CA
Arvixe Server Location: Santa Rosa, CA
Max Response Time 2.93 seconds
Minimum Response Time 1.06 seconds
Arvixe Server Location: Santa Rosa, CA
Max Response Time 2.93 seconds
Minimum Response Time 1.06 seconds
Dreamhost
While at a first Glance Dreamhost’s results may seem inconsistent, you’ll notice that there are more bumps in the road because the extremes are much more controlled. So variances in a handful of milliseconds show up as jumps in the graph. Overall Dreamhost was solid from beginning to end. It didn’t report the lowest lows, but it also kept things in check as traffic increased, without having massive jumps in response times. Dreamhost had a strong showing.
Testing Server Location: Ashburn, VA
Dreamhost Server Location: Los Angeles, CA
Max Response Time: 3.74 seconds
Minimum Response Time: 621.87 milliseconds
Dreamhost Server Location: Los Angeles, CA
Max Response Time: 3.74 seconds
Minimum Response Time: 621.87 milliseconds
GreenGeeks Hosting
GreenGeeks didn’t do badly at all in the performance testing. After a big initial spike in response time, the server settled down and returned the sub one second response times that we like to see. There was a bit more variance throughout the test than we normally like to see, but nothing that would indicate any type of major issue. For the most part things were pretty solid.
Testing Sever Location: Chicago, IL
Green Geeks Server Location: Dallas, TX
Max Response Time: 4.7 seconds
Minimum Response Time: 571.33 ms
Green Geeks Server Location: Dallas, TX
Max Response Time: 4.7 seconds
Minimum Response Time: 571.33 ms
GoDaddy
GoDaddy surprised me in more ways than one. GoDaddy started at a blazing 483 ms response time, but once traffic hit 25 users, it essentially fell off the face of the earth. The report lists times of above 4 minutes, and that may be true, but it almost looks like the server became completely unresponsive or started rejecting connections. The load test reported a number of failed attempts to connect to the server. While GoDaddy shined at lower traffic levels, it fell apart completely as traffic passed the 25 user mark.
Godaddy reached out and ask that I clarify the results of their test. This is what they had to say:”We use a software security layer called Sentinel. Because of it’s conservative settings, the software detected the load test as a DoS attack since all the LoadImpact traffic was coming from one IP, and banned it for 5 minutes.”I believe they have thresholds set at a painfully low level if they consider 25 users a DDOS attack. That means one small business sharing an article internally could take down a site.
Testing Server Location: Ashburn, VA
GoDaddy Server Location: Phoenix, AZ
Max Response Time: 4.1 minutes
Minimum Response Time: 483.08 milliseconds
GoDaddy Server Location: Phoenix, AZ
Max Response Time: 4.1 minutes
Minimum Response Time: 483.08 milliseconds
So which shared WordPress hosting company performed the fastest?
Based strictly on the performance data from each webhost, there were three hosts that really stood out in the group. Each of the following hosts had an average page load time of below one second throughout the entire course of the test, all the way to 50 concurrent users. Stay tuned for the next update in early 2014 where we re-evaluate all of these options.
I’d love to hear your thoughts on any of the data you see here. Anything that stands out to you? Anything that surprised you?
Let’s talk it out 
