Faster servers = better rankings? What to choose?

[Timetraveler]

From reading by Matt Cutts it's faster sites are better for seo as he says its one of their 200 signals (nothing new, he released it on his blog back in 2010).

Anyways, we're trying to speed up our sites and after a massive downtime issue with our current server company using both VPS's and a dedicated server I've decided its time to switch.

What I'm wondering is how do you determine what company to go with. Every site claims to be the best and fastest :(. Is there someway to determine what Google wants or what type of servers they prefer?

[Robert Charlton]

Note that we can't discuss specific hosting companies here (or in fact anywhere in the public areas of WebmasterWorld) but we can discuss the kinds of things to look for in a hosting company.

Regarding downtime... some years back, after WebmasterWorld had a week long outage because a backhoe had cut a fiber optic cable somewhere, I had a client set up failover DNS, with a backup server located in a geographically separated area. As it turns out, we've never needed it, but if you've got a business that's running on the web, it's a good option to consider.

Regarding site speed, in our recent SEO tools discussion...

Favorite SEO Tools
http://www.webmasterworld.com/google/4367011.htm [webmasterworld.com]

...one of the tools suggested was GTmetrix [gtmetrix.com...] which analyzes a site's speed and performance, and explains the factors which affect site speed.

I highly recommend the analyzer, and I think a discussion of some of those factors might be helpful.

PS: There is a Webhosting Issues and Options forum [webmasterworld.com] in the Supporters area where discussion of specific hosting companies is allowed.

[tedster]

In my experience, the server is not often the most important fix when it comes to site speed. Instead, it's text and image compression plus too many http calls first. Database optimization is sometimes in the mix at that point - but only when all that is addressed does server response truly enter into it.

I think it's a mistake to assume that "site speed" means "a faster server." It's essential to focus on the fact that this is a user experience metric! Along those lines, there is some JavaScript that Yahoo developed a while back called boomerang.js that lets you study your visitors' page processing experience directly.

Before choosing to upgrade your servers (and the value there is only a "maybe" as you've noticed) I'd put in a good study first of your page processing time.

There are free tools online that will show you how long it takes your server to respond to requests, as well as how long various page components take to render. You need that data if you're serious about improving your visitor's experience of site speed.

IMO site speed is a very good thing to be serious about! The gains are often a lot more than a possible uptick from the algorithm. I recently saw a large-scale study where a 1 second improvement in page processing time generated a 7% improvement in conversions - for an already highly successful major retailer. A related metric from the same study, a .5 second improvement in page processing speed generated a 20% increase in page views!

I have only seen one case where the server speed was the big issue. In that case there was no doubt about the fact that it was too slow - over 4 seconds for the first byte to be returned on-average! If the database isn't a horror show, any decent ISP can offer your better than that.

[Sgt_Kickaxe]

Since gtmetrix was mentioned above another popular and reputable free service is webpagetest.org and I suggest using both at the same time(and wpt from multiple countries/browser types), each have their pros and cons but combined you get an even clearer picture on some issues.

As for what to look for you'll need to look at the trickle charts either service provides to find the bottle necks.

- High TTFB(time to first byte) suggests server issues
- High image load times suggests needing a CDN(content delivery network) or image optimization/reduction
- Javascript placed before css will delay css load and cause a perceived delay for visitors(blocking).
- etc

Both sites have solid documentation on what you're looking at and WW is the place to ask questions :)

[Frost_Angel]

From a none-SEO expert website owners point of view - just thought I'd add this.

I too was heading down the path to switch servers - and I had just moved the site to a new server with all the bells and whistles a year ago. But I was reading all these stories about "changed my server and got out of Panda" and I was trigger happy after that. But I needed to STOP and think it through.

I decided to look closer at my site and dig deeper. Sure enough - lots of things to fix. Optimize images, put js in the footer, remove a video I had in the nav, scrutinized my plugins and removed what I "really" didn't need. A few other things as well.

It made a huge difference. I've been with my host since 1999 and I almost jumped ship. This host has gone above and beyond - and I almost threw it away because I was feeling "desperate" to get out of Panda. Desperation is not decision-making's friend. LOL

[robzilla]

I think the discussion title is misleading and incorrect. A faster server will not result in better rankings. What's probably true is that a slow server may have a negative impact on your rankings, as indeed it should, but beyond a certain point there's unlikely to be a direct advantage -- but by all means, improve your user experience.

[Robert Charlton]

I think the discussion title is misleading and incorrect. A faster server will not result in better rankings.

robzilla - Good point. Rather than keeping the OP's assumption as a statement, I've added a question mark after the first sentence, and for the record mention that here.

[netmeg]

I start with the platform I'm going to be working with, and work my way through that. Long and sometimes bitter experience, not to mention several years, before I found a host that really seemed to know how best to serve up WordPress (and it was my FIFTH host) If the host can't do decent job with the basic platform or CMS, before you've put in your plugins, your custom coding, whatever - the won't stand a chance. You should see how Magento (already slow) runs on a substandard host.

So when it came time to find the best host for myself or my clients, what I did was look around to see where some of my peers were hosting, or where my favorite well-performing sites were hosted.

It's also nice if you can find a host who will let you know when you add some stupid plugin or code that's gonna drag down your performance, too. (Yea, I've done that)