Once you’re set up on a web host, it’s easy to get complacent (“My site is up! I’m good to go!”). But there are several excellent reasons to pay attention to your host’s performance:
- It will definitely impact your visitors’ experience. You’ve heard it before but it’s worth repeating: people aren’t patient, and you’re easily replaced by another, faster search result. Even loyal fans will be annoyed when your site is slow. If you’re selling something, a slow site will lead to lost sales. If you’re blogging, you’ll get more bounces and less readers.
- It will probably impact whether people find you in the first place. Google has openly stated that they use site speed as one factor in how high a page is ranked. It’s certainly not the only factor, but all other things being equal, you’ll rank below a faster site.
- Slowness is an everyday hassle, and an outage is a major headache. When your site is slow, just logging in to create new content is a pain. And if your site goes down, suddenly you have a to-do-list-dominating priority, whether you have time for it or not.
Okay, got it. Slow site = bad. Now what?
If your site is really slow, you’ll know it already. But how can you evaluate your host if you’re not sure? That’s what the rest of this post is about. There are several areas that impact your site speed, so I’ll break it out into mini-tests you can do to check each factor.
Of course, your host isn’t the only thing that can affect your speed. Your site platform (I mostly work with WordPress), your theme, and your content will also make a huge difference. But testing your host is a good place to start, because it’s the foundation of the rest of the factors—even a beautifully optimized site will drag if its host is slow.
Note: These tests are for checking your existing host, and are targeted specifically at “shared hosting” accounts (because if you have your own server, I’m assuming you know a lot about these things already). I don’t know of any way to run all of them on a host where you don’t have an account, but many good hosts offer a money-back guarantee, so it’s something you can check with relatively low pain.
Within your web host, your site is physically on server somewhere. That server is connected to the internet in much the same way as your home computer is connected to the internet (except at your host, the connection is—hopefully!—much faster, and it’s always on).
The first three tests check whether there are any major speed bottlenecks just getting basic information to and from the host.
What you’re measuring: A “ping” is about the most basic kind of information you can transfer. The full process (in non-geeky terms!) of sending a ping is essentially one computer asking another computer, “Hey, are you there?” and the second one replying, “Yep, I’m here!” What ping speed measures is how long it takes for that message to be sent, arrive at the second computer, and the return reply to be sent and received.
How to test it: Just type in your website address over at check-host.net and push the “Ping” button. You’ll get back a list of results from a bunch of test locations around the world.
How to interpret the results: Ping speed can vary a lot around the world (because it’s directly related to physical distance as well as server and network speed), so we’re looking for trends. The number we’re looking at is the average number of milliseconds (abbreviated “avg”); the test sends four pings per test location, so using the average smoothes out any quirks.
Take a look at this test of www.google.com (I’ve highlighted the relevant numbers):
As you can see, most of the times reported are pretty similar (the average over all of them is 29.4 milliseconds, but you don’t need to do the math… we’re looking for rough numbers, and you could guess from glancing at it that the average is somewhere between 10 and 50 milliseconds).
So how fast is fast enough? Most people I asked said you want an average under 90 or 100 milliseconds. There’s no hard-and-fast rule; you’re mostly just checking that the results aren’t horrible. And pings aren’t the last word, either, just one hint. If you’re getting an average above, say, 200 milliseconds, it’s a warning sign, and if the other tests are also slow, it’s time to find another host.
File download test
What you’re measuring: Can the host serve files at a good speed?
How to test it: Upload a big file (50+ MB) onto your account at the host (using FTP, for instance) and then download it through the web and watch the download speed. (If you need a big file, you can find some here that are expressly available for testing download speed.) You don’t have to download the whole file necessarily, just watch it for a few minutes to get a sense for the speed you’re getting.
How to interpret the results: The major caveat on this test is that your own internet speed will affect the results. If the host is fully capable of sending the big file at a very fast speed, but your internet connection is only so-so, you’ll be limited by your connection. (If you’re not sure you have a good connection, test it at Speedtest.net. If that test reports your download speed to be at least 10Mbps, you should get valid results on the big file download test.)
As usual, “good” speed is subjective, but I’d look for 200 KB/second or more.
Simple webpage test
What you’re measuring: How fast does the host serve up web pages (the kind of content you’ll actually be hosting)?
How to interpret the results: If you’re using the page + image I provided, I’d expect the total time to be under a second (ideally well under a second). This is essentially your baseline. Your websites will only get slower from here, so make sure it’s something you can live with.
Just like the condition a physical neighborhood changes the value of individual houses, your “neighbors” on a hosting server can make it a lousy place to be. This can be active (people who use more than their fair share of resources or share files that trigger “objectionable content” filters) or passive (too many “average” people crammed onto one server).
I’m not aware of any perfect test for this, but there are two things you can check that will give you a general idea of whether you have a problem on your server.
Look for baddies
What you’re measuring: Are there sites hosted on your server that are making you guilty-by-association?
How to test it: Enter your website address in this Reverse IP Domain Check.
Is it worth moving hosts if you see this? It depends on your target audience and the results of the other tests. If you are trying to reach kids, students of any age, or a corporate audience, the real possibility of being blocked by filtering software could cripple your site. And if there are other concerning results, this would be enough to push me to the “move” side.
You may also have noticed the number of domains hosted on the server. It’s tricky to know what a good number is, because it depends on the quality and resources of the server, and that’s what the next test will tell you.
What you’re measuring: Is the server maxed out?
How to test it: It’s not free, but for $30, Status2K provides details about the server’s resources and how many of them are used. (If you want me to write a tutorial about how to install it, let me know.)
How to interpret the results: Here’s what my host looks like as I’m writing this:
You’re basically looking for long uptime, and a relatively small percentage used for each resource. I’m happy with what I’m seeing, because there’s enough unused resources in each area to handle any kind of normal traffic spikes.
If your host checks out on all of the above, the next step is to optimize your site. That’s a huge topic, but I’ll be tackling it in bits in future posts. If you have specific questions, post them as comments and I’ll update this post and write new ones as needed.