When determining the ranking of search engine results, Google considers a myriad of factors. Mostly, these factors concern webpage content, the trust factor of the URL, titles and headers and the age of the domain name.
In 2010, Matt Cutts announced that the website speed of a site would begin to have an impact on ranking. He went on to confirm later that year that mobile sites with a slow performance would suffer a penalty in search rankings. Because a slow website results in a poor user experience, the slower the site, the lower the ranking.
Should you be worried?
If you’ve paid for hosting on a shared server, you’re probably already all too aware of slow site speeds every now and again. On average, users of a website will leave if it hasn’t loaded in ten seconds – high bounce rates on a home page should be enough to tell you there’s something going wrong.
Take a look at your main landing pages. Are they packed with images? Are there lots of videos, or a whole pack of adverts? Especially when accessed on a mobile, your site is going to take forever to load. And that isn’t going to be good for rankings.
That said, Google is not especially clear on exactly what denotes a page load time. Does it mean document complete time, or fully rendered time?
Moz, which is an SEO site you’ll probably already be very aware of (expect me to reference this a lot, since I’m pretty much in love with Rand Fishkin) did a test to check if full render times and document complete times had much of a difference in search rankings. The results were pretty interesting – well, pretty interesting to someone as worryingly interested in SEO as you and I:
As you can see, the differences between the two factors is minimal – even though full render time takes longer, there’s no difference in search rankings. Whether this is something which will be changed in the next Penguin version is unclear at this point.
However, further analysis showed that pages with shorter time to process the first byte (TTFB) by the web server were consistently higher in search rankings.
So it’s the bytes which matter in page speed ranking factors.
Depending on the kind of back-end performance your website has, that crucial first byte will be delivered at a certain speed. And by back-end I mean back-end applications, database servers, network connections and web servers.
Moz suggest looking into optimising application codes and database queries after checking your TTFB with WebPageTest.
I’d also recommend taking a good look at your web servers. Research has shown that users will leave a site if it hasn’t loaded in ten seconds, so choosing a hosting provider which can handle whatever you throw into the site is pretty essential.
One example which springs to mind is my procrastination-enabling informative website Whimsysaur, which has all the hallmarks of being a pretty poor site from an SEO perspective. It doesn’t rank for any search term and, for reasons I have yet to understand, has a PR of 0.
Pages on Whimsysaur routinely take more than ten seconds to load. This is probably to do with my cheap hosting, as it appears to have good days and bad days. In any case, it accounts for the embarrassingly high bounce rate on landing pages. Aside from the fact that the site doesn’t look particularly good, I mean.
Of course plenty of things affect the search rankings of your site; and your TTFB is just another one of those things. Essentially, Google aims to get content and sites which actually garner a good user experience ranking high on a search – so if you’re not getting much traffic, you need to work out why.
I’ll do a quick overview of how to pinpoint key issues with your site using Google Analytics in a post soon!