How to Choose the Best WordPress Host for X Theme

As hosting is always an ongoing discussion in the X Theme and Pro Theme community, we decided to put the most popular host recommendations to the test. We scrutinised them for speed and capability to handle large numbers of visitors to the site. We also measured benchmarks for how quickly they can run the code that makes up WordPress websites.

The goal here is to help you, the general user, understand some of your hosting options and to show you some real world results you can expect from them. To do this, we tested specifically with an actual working X Theme site, but these tests will essentially hold true for most themes, especially other all-in-one themes like Avada, Divi and more. In most cases, we aimed for intermediate hosting plans that were around the $30 per month mark. These are better (in terms of performance) than your average entry-level starter plan, but still quite affordable for any business. The better performance came in the form of increased resources such as RAM and CPU Cores. The small increase in resources gives a large boost in the server’s ability to handle traffic spikes, something that all website owners should keep in mind.

We have broken this page down into these parts:

• Part 1: Hosting 101
• Part 2: Hosts and Plans Tested
• Part 3: Tests Performed
• Part 4: Speed Test Results
• Part 5: Load Test Results
• Part 6: Benchmark Results
• Part 7: Individual Host Summaries
• Part 8: Conclusion & Our Recommendations

So how do you choose the best WordPress host for X Theme? You’d better strap in — this is a long one! But definitely worth the read as you’ll come away with a better understanding of how hosting works and how to make the best choice for your site.

For many people, trying to figure out the difference between hosting packages is confusing and feels like it requires a PhD in rocket science and maybe minor in neuroscience! What is Shared Hosting? How about Cloud Hosting? How much RAM do I need?

Shared Hosting is the cheapest, most basic hosting that you can get. Most people new to hosting select a shared hosting package without understanding it. Shared hosting is cheap,well-advertised, and because of that, it seems appealing. It is cheap though for a very good reason — most shared hosts stuff tonnes of sites onto the server, filling it up and using all the resources. This can result in slow sites and errors. It may have perks in the sense of available resources for traffic spikes, but few hosts provide this efficiently. Shared hosting also has a higher risk of security vulnerabilities due to server wide compromise. Let’s be super clear — neither Michael nor Roy recommend shared hosting very often.

A VPS is where one big, super powerful server is divided up into small virtual servers. Each one has a set portion of resources allocated to it, usually with its own IP address etc. With a VPS server, if you pay for 4GB of RAM you always have 4GB of RAM available even if you only need 1GB. One drawback to this approach is that if you happen to need more than 4GB, (due to an unusual surge in traffic, for example) you may run into trouble. There simply isn’t any more memory available for your site to use. Some specialised VPS providers provide auto-scaling – they track the resources needed and try to automatically add more when required. We’ll be testing one such example from Kyup below. In most cases you can also track the resources used by your server and scale up. However, this usually doesn’t happen until after users start experiencing errors and is often a permanent change.

Being 100% honest — many believe the term Cloud Hosting is just marketing. In theory, Cloud Hosting should be a service where you can tap into resources as you need, and some of the more sophisticated Cloud hosts work this way. But in reality most that you see are structured more like a VPS – you get a certain amount of RAM, 1, 2 or more CPUs and so on. Scalability is the biggest selling point of cloud hosting, as it allows for infinitely scalable resources when needed. That means if you suddenly get an influx of traffic, your site won’t go offline. However, not all hosts handle scaling well, and some may impose some limits on scaling in order to avoid certain clients using a high amount of resources and abusing the system.

Managed WordPress Hosting is a service rather than a specific type of server. It can apply to shared, VPS, cloud and dedicated hosting plans. Managed WordPress Hosting adds a premium to the cost of your hosting due to the level of service you are provided. As a user that doesn’t want to manage as many of the technical aspects of hosting, it can be an incredible time saver and is usually a better option than the others.

With Managed Hosting the hosting company will install WordPress, update the core WP files and often provide server-level caching, backups, and a staging version of the site. They’ll also generally handle many of the hosting side issues that come up and may provide more in-depth WordPress-specific support as well. On a non-shared environment, the managed hosting company will also take care of the server’s software and security.

A few notes on the hosts and plans selected:

Shared Hosting: We only selected Siteground. There are a couple of reasons for this. First, we don’t really recommend shared hosting so we didn’t want to test a bunch. Second, we picked the one that seems to be the most popular and frequently recommended.

Packages: We tried to pick packages and options that were similar. We didn’t throw WordPress VIP (starting at $5,000/month) against the cheapest Shared Hosting from GoDaddy. The goal was to keep things relatively similar in terms of packages to show off the differences in hosts. The only package that stands out is Kyup, and that is because all of their packages are more expensive. Our goal was to try to use a 2GB RAM server in all instances. Important to note is that Linode is completely unmanaged.

Tested Site: Instead of using a base theme or default WordPress 2017 theme, we decided that a real site with real content, as well as the normal optimisations (or lack thereof) would be best. That’s why we call it a real world test, as most of you will also have contact forms and plugins running. We used an identical clone of our own current X Theme Tips website for all testing. The only difference is where a host suggested disabling plugin based caching. In those cases, we turned off WP Rocket and cleared out .htaccess, etc. There will be relevant notes on this.

Features: Almost all hosts offered PHP 7 and all offered free Let’s Encrypt SSL. We enabled those options. Where possible we enabled performance options provided by the host such as MariaDB, server caching and so on.

Plugins: We used the same plugins as the site uses everyday. This includes WP Rocket (except for WP Engine)

Location: Where possible for each host, we choose a data center near New York. The only cases we didn’t were where the host didn’t offer any choice/control.

All prices in USD

If you know anything about hosting, you know that a good host performs well in several areas. We selected a comprehensive series of tests to try and hit the various areas. We wanted to know how well the hosts executes PHP and SQL, how fast it could serve pages (Time to First Byte, page load, etc.) and how it performed under load. This multi-pronged approach helped to determine which hosts performed well overall, versus hosts that performed well in certain areas.

Speed Tests

For the speed tests, we used Pingdom Tools, GT Metrix and Web Page Test. The same settings and location were used for all tests. Here we are testing the time to first byte and page load speed. We were not worried about the Pagespeed or ySlow score, but solely about how quickly the page would load. All tests were run unthrottled. While they won’t directly reflect real users’ experience, this eliminates variables, making it more reliable to compare tests within each tool. We stress again that these tools are great for a comparative analysis amongst themselves, but not against each other, as you will see in the results.

Load Tests

A load test simulates users accessing the site. All of our load tests had the simulated user accessing the same two posts (we selected the two with the largest number of requests and largest physical size). The tests were 250, 500, 1000, 1500, 2000, 2500, 5000 and 10,000 users accessing the site over 1 minute. We also did what we call the insanity tests, where we tested the servers with a level of visitors similar to the “Slashdot Effect”. Those tests were 250, 500, 1000, 1500, 2000 and 2500 users accessing the site per second. We load tested the sites on two separate days for verification to ensure there were no anomalies during the first batch.

PHP and SQL Benchmark Tests

We used two scripts. The first one tested the performance of PHP on the server. The second script was more reflective of the typical WordPress environment, in that it tested PHP and SQL execution.

You can check out the scripts with the below links:

• PHP Benchmark Script
• PHP and SQL Benchmark Script

This is the real bread and butter comparison that most people are looking for. As we mentioned earlier, these are real world results. What we mean by that is you can expect to see similar results with your WordPress or X Theme site. We took a backup of our very own X Theme Tips website and cloned it to every single host, keeping each one configured as close to identically as the host would allow. Could we have used a more optimized site? Of course. But not many of you are running the default 2017 theme with no images or plugins, so what good would that do?

Speed Tests – Pingdom

Pingdom usually reports the fastest load time for a page out of the three test sites we used. This is a widely known truth and the reason why it’s the most common result shared when people talk about their site’s load speed.

Results (time in seconds)

1. 0.768 Linode
2. 0.842 Amazon Web Services
3. 0.864 Vultr
4. 0.964 Siteground
5. 1.020 Kyup
6. 1.180 Google Cloud
7. 1.180 WP Engine
8. 1.290 Digital Ocean
9. 1.790 Themeco
10. 1.830 Flywheel
11. 2.230 Atlantic

Speed Tests – GT Metrix

For this test, we enabled the stop onload feature to better simulate real world experience. You will need to sign up for a free account in order to do this. One perk of that is the free monitoring feature, where you can have your site scanned at specific intervals for your reference.

Results (time in seconds)

1. 0.9 Vultr
2. 1.3 Linode
3. 1.4 Siteground
4. 1.5 Digital Ocean
5. 1.7 Flywheel
6. 1.7 Kyup
7. 1.8 Themeco
8. 1.9 Amazon Web Services
9. 1.9 WP Engine
10. 2.5 Atlantic
11. 3.3 Google Cloud

Speed Tests – Web Page Test

Web Page Test allows you to run the same test numerous times and gives you a “first view” and “repeat view” time on each. We recorded the first view time to get a better feel for real world performance of your site for a first time visitor.

Results (time in seconds)

1. 1.932 Linode
2. 1.959 Vultr
3. 1.998 WP Engine
4. 2.178 Amazon Web Services
5. 2.264 Flywheel
6. 2.486 Digital Ocean
7. 2.587 Siteground
8. 2.613 Google Cloud
9. 2.627 Kyup
10. 3.074 Themeco
11. 3.507 Atlantic

Summary

If speed is your overall goal, then Cloudways Vultr cleaned up in this category. Linode was an extremely close second place, followed by Cloudways AWS.

It’s important to note that the size of your site and how far you’ve gone to optimise it will have drastic effects on your load times. And while using one of the winners above will give you a large head start, it pays to go through your site and reduce inefficiencies as needed. All three of the tests above will give you recommendations on how to do exactly that.

Load tests are realistically a far better measure of a server’s performance, as they will tell you exactly what happens when a certain number of visitors are simultaneously visiting your site. In the tests below, the greyed out “failed” bars represent tests that either timed out or had error rates (the % of hits that will receive a server error instead of a page) above 10%.

We called the next tests Insanity Tests because, to be frank, this level of traffic is rare for most websites. But it can happen. Have a post or product that goes viral? Boom, thousands of hits at once. Can your server handle it?

The first thing you will notice is a lot of gray bars. As above, the grayed out bars are tests that failed. You may also notice Themeco has none. That’s because it was the only host we tested that didn’t fail a single load test.

Below is a breakdown of each host’s Insanity Test results. As most did well in the normal load test, we decided this breakdown was the best to show. The orange area fill is the error rate, and the grey bars is the average response time of each request at that specified load target.

PHP Benchmark

Let’s start with the biggest losers (with good reason) here. Themeco and Flywheel trailed significantly behind the others and this was no surprise. These hosts were running versions of PHP 5.6 which is slower than the PHP 7 version other hosts were running. When these hosts can offer PHP 7 we would expect to see the results improve significantly.

1. Google Cloud
2. Amazon Web Services
3. Kyup
4. Vultr
5. Siteground
6. Digital Ocean
7. Linode
8. WP Engine
9. Atlantic
10. Flywheel
11. Themeco

In terms of winners, we see the ‘enterprise level’ solutions winning, with Google Cloud the fastest, Amazon Web Services second and Kyup third. After that, we see Vultr, Siteground, Digital Ocean and Linode all grouped pretty close together.

We didn’t factor the results of this test into our overall recommendations solely due to the reason that WordPress rarely relies on PHP computation without MySQL. It does, however, show the variance in processing power between similar setups.

PHP + SQL Benchmark

This test was actually really interesting and reflects the importance of considering the performance of PHP and SQL combined which is the cornerstone of WordPress. If you have a query-heavy WordPress site, this test will be of more importance to you.

1. Google Cloud
2. Kyup
3. Linode
4. Digital Ocean
5. Amazon Web Services
6. Siteground
7. WP Engine
8. Vultr
9. Themeco
10. Flywheel
11. Atlantic

The winner in this test was again Google, followed by Kyup, Linode, Digital Ocean and AWS behind them.

As you can see most of the hosts were similar, however, if you have a large site with complicated queries, you may wish to focus on the results above with the lowest computation times or choose a VPS option where you can scale the server as needed.

Amazon Web Services coming in at number 1 isn’t surprising. They are the big boys on the block when it comes to hosting. If you are willing to invest the money, they are one of the best options on the market.

Pros: When choosing AWS through Cloudways, you can easily leverage the power of AWS without the setup headaches. Going through AWS directly will require quite a bit of technical knowledge.

Cons: You start with 2GB of Bandwidth, 2GB of storage and 2GB of Database space. When it comes to storage, this isn’t particularly a tonne of space. If you are planning to run multiple sites off of 1 server, and end up needing space the price will go up. Bandwidth would be the most costly addition. However, as you increase your storage space, database size and bandwidth your costs will go up. The jumps in server tiers (CPU, RAM) is very expensive.

Overall Vultr performed very well and might be the clear cut winner of the tests when all things are considered. You can run a number of sites on one server and it will handle the load and run fast. Using Vultr through Cloudways makes it easy to host your sites, or host a number of client sites.

Pros: Fast. Did we mention fast? Vultr beat everyone, including AWS. It came second the load testing and performed very well in general.

Cons: It isn’t quite as easy to use as WP Engine or Siteground, so you may need to be a little more comfortable with it in order to use it with your site or your client’s sites. Cloudways does have an intuitive UI however, making the learning curve easy to master.

Note: This particular plan is no longer available, but its replacement includes even more resources so performance will almost certainly be higher.

Linode and Vultr are, despite AWS ranking first, the two winners of this test when considering the market the testing is aimed at. They were top 3 in all areas tested, with blazing fast loads times.

Pros: Fast. Great in load testing. Affordable. Multiple add-ons available for users (at extra cost), such a load balancing.

Cons: Linode can be a little harder to manage than Cloudways. We recommend a system like Server Pilot to make it easier. You won’t have the support you do with Cloudways, so definitely understand what you’re getting into before opting to go with Linode. It is not a turnkey solution.

Digital Ocean is a crowd favourite and these tests show why. If it wasn’t for Vultr and Linode, Digital Ocean would have been second behind Amazon Web Services. Digital Ocean ranked well across the board. While it didn’t smoke any tests, it performed very well in every category.

Pros: Solid price point. You can use it with Cloudways as we have, or with Server Pilot or as a true VPS (the latter two may perform differently as they probably won’t have the setup Cloudways does).

Cons: Given the overall package, there aren’t any major cons. It could be a little faster and handle load a little better, but Digital Ocean is no slouch.

(Update summer 2018: Themeco has discontinued their hosting service.)
Themeco hosting provided us with some challenges. First, they were the only host where we didn’t move the WordPress site ourselves – UpDraftPlus migration failed on moving/unpacking the plugins directory. We didn’t troubleshoot the problem (directory permissions? server restrictions? Not sure) but instead had Themeco migrate the site for us. Our first set of tests were poor, but we reached out to Rubin and Kyle from Themeco and found out there were some technical problems at the time of the testing. Once they were done we began retesting. We then discovered that during the migration, the .htaccess file had caching related directives left over from the previous install that were impacting performance. After some help from Rubin to pinpoint that issue, we were off the races. Interestingly, we found that our tests were far better with WP Rocket enabled than without (we note this as Themeco generally recommends turning off a caching plugin). That said, we made sure to clean .htaccess out of caching rules and then enable WP Rocket so the rules added were fresh. Moving forward, we would recommend that you allow them to migrate your site for you (free of charge) to avoid any unforeseen issues.

Pros: Not only the best host on high load testing but the only host to actually pass our insanity tests. Themeco slowed down, but the error rate never went over 1.9% on the tests. We also ran 5000 users/second and 7,500 users/second and the response times and error rates were 6225ms and 6.0% and 9670ms and 9.1% respectively. While your site will slow down if the load is extremely high, it will still load.

Cons: Themeco was one of the slower hosts tested, ranking 9th overall on the speed tests. They were however still within range of the others, so it may not be a deal breaker for all users.

WP Engine actually didn’t perform quite as well as expected. They are a premium host and their performance was okay, but not great. One huge concern for existing WP Engine customers running PHP 5.6 is the upgrade process. Most hosts (and all the other hosts tested) either come with PHP 7 or offer an ‘upgrade in place’ where the existing server or hosting account is upgraded. With WP Engine you must migrate to a new account and your IP address will change. This could cause some disruption if you’re running an e-commerce store, community site or something with constantly updating/changing content.

Pros: Like Flywheel, WP Engine is a fully managed WordPress host. They offer easy site staging, and give a little more control than Flywheel through their dashboard.

Cons: Given the price point, you’re not getting a great deal performance wise. They are solid and reliable and known to have great support, but the performance is nothing to blow your socks off.

The 7th place rank of Google Cloud was, to be honest, stunning. (As a note, it’s possible this was something with the Cloudways stack/setup). Their slow speed, in particular, was a big surprise. Google can clearly deliver content fast and effectively, but that was not on display with their hosting package.

Pros: Their PHP and PHP+SQL benchmarks were fast, so if your site is going to be processing lots of queries then the server should be able to handle that better than other hosts.

Cons: At a price point close to AWS, there doesn’t seem to be much reason to pick this option. We would take Cloudways/AWS over Cloudways/Google Cloud Platform in a heartbeat.

Siteground was our only shared host, but they did decently well. Their speeds are good, but they definitely suffered on the load testing.

Pros: Simpler hosting to set up and start with, and lots of support.

Cons: If your site is going to have much traffic, it will definitely affect page load times and performance. When you consider that we tested with the GoGeek package and they list it as good for about 100,000 visits monthly our test results agree. 100,000/month is about 3,333/day and 139/hour, or basically 2.5/minute. If you hit these levels consistently, it is reported that Siteground will ask you to upgrade to their cloud platform, which has always performed worse in any test we have seen or done ourselves.

Atlantic was a bit of a dark horse. It was offered up to us free by an X Theme Users member. This is a true VPS, where you will need to understand Linux servers, web servers and such in order to manage. You will start with a clean slate and use server administration tools and software to set up all aspects of the operating system, web server, security etc. you want. The performance of the server is truly in your hands, but it will take an experienced user a few hours to set it up. An inexperienced user may take days.

Pros: We don’t want to pick on a host, but there isn’t a lot to recommend with Atlantic. At the price point, we tested a few other great VPS choices. It did perform moderately well with our load tests, coming in around the middle of the pack.

Cons: Atlantic was the slowest of all the services tested, and given the value being placed on site speed this is a huge drawback. To run a site or sites on this host also requires a lot more work than most hosts, so between the work involved and slow sites we are not recommending Atlantic.

The biggest surprise with Flywheel is that they don’t offer PHP 7 yet (at least not at the time of testing). We think that this was a key factor in their performance. Overall getting a site online with Flywheel is really easy and fast, and their interface is pretty simple and straightforward.

Pros: Flywheel is a fully managed WordPress host. Their system and support will help someone who doesn’t know anything about hosting run a WordPress site. They offer easy staging, so you can design a site in private, or apply and test changes or upgrades before going live. While they might not perform as well as WP Engine, their price point might make them more appealing.

Cons: If you want any server side control, you won’t get that from Flywheel. This is the flip side of managed hosting. Overall their benchmarks, site speed and load capabilities were subpar compared to other hosts in the test.

Kyup is an interesting host for one very particular reason. It is one of the few to offer auto-scaling, and the only one from Cloudways to do so. While it is easy to scale a VPS up, it is often not possible to scale it down (shrinking the storage space is the problem). Kyup can auto-scale up for heavy loads, and then scale back down when the load returns to normal.

Pros: The only pro, from the testing, would be auto-scaling. If you expect your site to occasionally get hit with very high traffic loads, Kyup is a host to consider.

Cons: You will pay and pay dearly to run Kyup with auto-scaling. Kyup was at least twice the price of any host we tested and considering they came dead last you seem to be paying that price for just auto-scaling. While their benchmarks for PHP and PHP+SQL were solid, they were slow and the least effective host at handling the load tests.

So what is the best WordPress hosting for X Theme? Our answer is: it depends. One has to take into account their goals, capabilities, and their budget, rather than just raw speed. Do you need a turnkey solution? Are you willing to get your hands a little dirty with server tweaks? Is the hosting plan for you or a client? Will you host multiple sites?

There are many variables in a real world test and just as many in a real world decision. However, we make the following recommendations:

Best WP Host for Someone Non-Technical:

WP Engine
While WP Engine performance was middle of the pack, it offers someone new to running a site lots of support and easy to use tools. You don’t need to understand much about hosting to get your site humming along. Their support is great for people new to their own site. Check out WP Engine
Special offer: get 4 free months with a WP Engine Annual plan

Best WordPress hosts for someone wanting a level of server control:

Cloudways and Linode
Depending on your technical knowledge, both Cloudways Vultr and Linode will offer fast servers that can handle user load really well. You can easily host several sites on one server and not max out resources. You could pay about $35/month and easily host a few dozen small business sites. We are recommending Cloudways Vultr as you get a managed host with technical support and a solid performance for your site. If you are designing sites for clients, hosting sites and looking for ways to easily create staging sites/development sites, then Cloudways does all of that and more.

This site is hosted on Cloudways Vultr.

Check Out Cloudways Exclusive Promo! Use coupon code XPROS for a $30 credit to Cloudways, courtesy of Michael Bourne!

Check out Linode

Know someone who is considering new hosting for their WordPress website and would appreciate this information? Please share these tests using one of the buttons below.

And if you have questions or feedback, we’d love to hear from you in the comments section.