There has been a lot of discussions, and even experiments to try and determine the extent Page Speed is a Google Ranking Factor. Unfortunately, for the most part, it would appear that only very slow websites are actually penalized.
On a side note, comments have been made by Google representatives that they are looking into making mobile page load times a ranking factor, but this has not been implemented yet. We will discuss all the details in this article.
To gain a sense of the potential impact, we should start at the beginning with the original 2010 post by Google and the comments leading up to that post by Matt Cutts. We can then look at some of the tests carried out by SEO specialists, and then finally comments by Google spokespersons about the issue.
We will then discuss more recent developments relating mobile browsing.
Matt Cutt's First Hints — 2009 \ 2010
While the official announcement post didn't arrive until a few months later, Matt Cutts started hinting and explaining the reasons why a faster website may rank higher in the SERPs.
The comments by Matt Cutts was an interview he carried out in November 2009 with WepProNews at PubCon Las Vegas:
We are starting to think more and more about should speed be a factor in Google's Rankings. Because even in Adwords if your website is slow that can be a factor in how much you have to pay in Adwords. Historically we haven't used it in our search rankings but a lot of people within Google think that the web should be fast. It should be a good experience and so it's sort of fair to say if you're a fast website maybe you should get a little bit of a bonus or if you have an awfully slow website maybe users don't want that as much.
This is quite interesting as it is one of the few comments made by Google or their representatives that indicate a positive ranking boost for the faster websites, rather than just a decrease in ranking for the very slow ones.
The next comment made by Matt Cutts was in a direct response to the following question on February 1, 2010:
Since we're hearing a lot of talk about the implications of Page Speed, I wonder if Google still cares as much about relevancy? Or are the recentness and page load time more important?
This is interesting as it endeavors to find out the scale or weight giving to page speed as a ranking factor. Here is what Matt Cutts had to say in response:
Now, relevancy is the most important. If you have two websites that are equally relevant same back links, everything else is the same, you'd probably prefer the one that's a little bit faster. So Page Speed can, in theory, be an interesting idea to try out for a factoring and scoring different websites. But absolutely, relevance is the primary component and we have over 200 signals in our scoring to try to return the most relevant, the most useful, the most accurate search results that we can find.
So, that's not going to change. Our philosophy will always try to be "Try to return the best page to users." Try to give them information about what they typed in.
But if you can speed your website up, it's really good for users as well as, potentially down the road, being good for search engines. So, it's something that people within Google have thought about. Wouldn't it be great if we could find a way where websites that are especially fast, or websites that are really, really, really bad experience for users.
It would be only one of over 200 different factors, but still could be a small factor in saying, "Yeah, you know what? This person's worked hard to deliver a really good user experience in addition to be in relevant." So, let's show that to users.
Again, this looks at both a positive boost for the faster websites, and a penalty for the slower ones. You can watch the full video below:
The Original Google Blog Post — 2010
While there had been previous mentions about website speed becoming a ranking factor, it wasn't until Google set out in a blog post entitled "Using website speed in web search ranking", dated April 9, 2010, that we received confirmation that it had actually been implemented:
As part of that effort, today we're including a new signal in our search ranking algorithms: website speed. Website speed reflects how quickly a website responds to web requests.
The terminology used here is quite interesting. They could have said "the time the website takes to load", but they chose the wording carefully to indicate the response to the request. We suspect from this that they may just be recording the time to first byte, i.e. the time it takes for your browser to start receiving information from the server after first making the request from the browser. This would make sense as it would be something their web crawler could easily measure at scale. We revisit this point further in the next section.
In the original post, Google quoted internal studies confirming research that shows that when a website responds slowly, visitors will spend less time on it. In addition, faster websites can reduce operating costs.
They state that they use a variety of sources to determine the speed of a website relative to other websites, and then list some free tools that can be used to evaluate website speed. You can view our detailed and more up to date post on measuring the speed of a website here. In that post we also discover why speed test results should not be interpreted too literally, as some measures to speed up your website can create false negative flags on some of these tools.
Matt Cutts chimes in
On the same day the official blog post was released, Matt Cutts also made a post on his blog discussing the matter further. The full post by Matt Cutts makes quite interesting readings about the motives behind the change, but the following points are the most significant:
I would love if SEOs dive into improving website speed, because (unlike a few facets of SEO) decreasing the latency of a website is something that is easily measurable and controllable.
The specific mention of latency further enforces that they are not measuring the total load time on the page, but instead the time taken for the initial response.
Matt Cutts ends the post with three take-away messages, including the fact the impact of the change is small:
First, this is actually a relatively small-impact change, so you don’t need to panic.
Second, speeding up your website is a great thing to do in general. Visitors to your website will be happier (and might convert more or use your website more), and a faster web will be better for all.
Third, this change highlights that there are very constructive things that can directly improve your website’s user experience.
The Moz Speed Ranking Tests — 2013
The next piece of the puzzle came thanks to a set of speed vs ranking tests carried out by Billy Hoffman in a post on Moz dated August 1, 2013. He was helped by Moz Data Scientist Matt Peters, and Zoompf.
Hoffman tested over 40 different page metrics for 100,000 different web pages, so as you can appreciate, the scale of this test was significant. The full data from this test can be downloaded here.
According to Hoffman, while they tested over 40 different metrics, the majority of them had no influence on the rankings. One example given, was that the number of connections a web browser uses to load the page had no relevance.
Here is a summary of the results:
1. Median Page Load Time
They tested both the "document complete" time and "fully rendered" time. They found no correlation between the complete load time of the webpage and rankings.
2. Time to First Byte
As we have mentioned a few times in this article, they then measured the Time to First Byte, or initial response time for the browser to receive the initial response from the server.
Remarkably, this did show a very small correlation.
The results make perfect sense. Time to First Byte is the easiest, and simplest metric for Google to capture, as it can be done by their existing crawlers.
Hoffman points out that the Time to First Byte is east to calculate and is a "reasonable metric to gauge the performance of an entire website". He states that it is affected by three factors (we have added some simple ways to improve them):
- The network latency between a visitor and the server — You can improve this by choosing a web server close to your visitors, or using something like CloudFlare Page Rules to cache your content and serve it close to your visitors.
- How heavily loaded the web server is — This is easy. Just choose a great web hosting provider.
- How quickly the website's back end can generate the content — This can be solved with page caching. Check out our W3 Total Cache & CloudFlare tutorial to gain significant speed improvements.
Not all experts agree that Time to First Byte is a useful metric. Just take CloudFlare, whose performance optimization focuses more on the experience received by end users, rather than a specific metric. As you will read later, maybe Google representatives agree with this statement.
Is the Ranking Factor being rolled back?
Now that we have a good grasp of previous developments, lets fast forward to 2015 and 2016 and see what Google Webmaster Trends Analyst, John Mueller, has to say.
Google Webmaster Central Office-hours Hangout — April 2015
This is the first indication we have from Google that website Speed is not being used much as a ranking factor. John Mueller, Webmaster Trends Analyst, said:
I don't know how much of that is still used at the moment. So we do say we have a small factor in there for pages that are really slow to load where we take that into account. But I don't know how much that's actually still actually still a problem in ranking.
It clearly indicates that there is no positive ranking factor for fast loading websites, but that they may penalize very slow websites by a very small amount.
You can watch the relevant part of the video below:
John Mueller on Twitter — November 2016
Fast forward to just a few days ago, and John Mueller gave some concrete advice for what you should target for your website load time.
Just looking at the two most recent comments by John Mueller, it is arguable whether the ranking factor even exists anymore. It has even been suggested by some that if your website is too slow, the crawlers will not even index it.
The Crawl Budget
What we do know is that Google's assigns a crawl budget to your website, which according to Matt Cutts in 2010 was "roughly proportional to your PageRank", but he did stress it was not a hard limit. It can also be impacted by the host load:
By far, the vast majority of websites are in the first realm, where PageRank plus other factors determines how deep we’ll go within a website. It is possible that host load can impact a website as well, however.
So, if you have a slow website, you may find less pages are crawled, and this may reduce the amount of content Google picks up from your website, and affect rankings that way. But this is pushing the boundaries of the discussion a little.
Mobile Speed as a Ranking Factor
Google Webmaster Trends Analyst, Gary Illyes, indicated in May 2015 at SMX Sydney that they were looking at ways to make Page Speed become part of the mobile-friendly signal group:
In further clarification of the efforts toward this end, Gary Illyes responded to a tweet in June 2016 confirming that a mobile page speed factor was still planned, but they had nothing to say just yet:
@jenstar for now this is still just a plan so there's not much we can say, but generally aim for great scores on the pagespeed testing tool— Gary Illyes ᕕ( ᐛ )ᕗ (@methode) June 1, 2016
When it comes to mobile search, you may also want to read up on the following:
Unfortunately, we do not have a 100 percent guaranteed answer as to whether Page Speed is a Google Ranking Factor at the present time, but we suspect there may be a small penalty if your website is extremely slow.
In 2010, it was clear Page Speed was effected rankings both positively and negatively, depending on which end of the spectrum you were on. Then, in mid-2015 John Mueller raised doubts as to whether that part of the algorithm was even used anymore although admitted it may still apply to really slow websites.
More recently John Mueller recommended a load time of under 2-3 seconds, but no mention was made as to whether that was just for user satisfaction, or for their algorithm.
Despite an apparent reversal of page speed being used for determining rankings, more recent comments have indicated a return of the ranking factor for the movable search index, although this seems to be in its very early stages.
On a slight tangent, Google confirms that it is giving higher rankings to mobile-friendly websites (which often are built to load faster on mobile), and more visibility to Accelerated Mobile Pages (Fast labels, exclusive carousels) which are highly optimized for speed and cached directly by Google around the world.
It is clear website speed it going to continue to play a factor in some form or another, even if just in relation to mobile websites.
Regardless of the above, there are many reasons from a conversion or user retention perspective to ensuring your website is fast, which we have gone into some detail here.