Blog

Does your load time affect your search results? Oban’s most recent case study.

When Google originally announced that page loading speed had become a factor in their algorithm in April 2011 initially webmasters and SEO’s did not believe it would become a significant ranking factor. However, in the time since, it has become apparent that Google wants to employ these usability metrics to help rank pages. It makes sense – how many times have you become frustrated and left a site because of slow loading speeds in these high speed broadband days? Furthermore, if it’s a slow ecommerce site, webmasters are effectively throwing away money as well as traffic.

So just how important is load time? If we look at Google’s official blog post announcing site speed as a factor, it says:

“While site speed is a new signal, it doesn’t carry as much weight as the relevance of a page. Currently, less than 1% of search queries are affected by the site speed signal”

Geoff Kenyon of Distilled posted his page speed experiment on the SEO Moz blog in February 2011 and concluded:

“(I think) this means that site speed will affect only queries where other ranking signals are very close or when the load time is exceptionally poor. If competing pages have high relevancy scores and close link metrics (which isn’t probable) then page speed may come into play. Additionally, I believe that site speed could negatively hurt you if your page takes an excruciatingly painful amount of time to load.”

We encountered the aforementioned very slow page load times with one of our clients and their website developers earlier this year. The case below outlines the process that we went through to find a solution for this issue and how we advised the web developers of the relevant changes they needed to action to decrease the load time. The following provides evidence that supports page loading time as a key ranking metric in Google:

At the end of March 2011, 90% of the rankings for our target keywords dropped out of the top 200 having previously been sitting pretty on and around page 2 for a number of months. We had initially attributed the drop in rankings to a large number of broken external links pointing to the site so we instructed the site’s developers to implement 301 redirects on the 404 pages to solve this issue.

Some of the rankings had begun to rise following the implementation of the necessary 301 redirects across the site. However, it was proving to be a slow process and there was still a lot of fluctuation happening, sites dropping in and out of the top 200 frequently. The majority of the rankings were not breaking in to the top 100 as expected.

After a couple of weeks the rankings had not returned – despite our confidence that they would. We compiled a site speed test and identified that the site was loading incredibly slowly (in reality, even slower than the tests below demonstrate). A slow server response time was identified as the main issue with the site taking around 11 seconds to even start downloading any content.

Here’s what Google page speed test had to say about the site on the 15th of May:

“On average, pages in your site take 10.5 SECONDS TO LOAD (updated on May 15, 2011). This is SLOWER THAN 95% OF SITES. These estimates are of LOW ACCURACY (fewer than 100 data points). The chart below shows how your site’s average page load time has changed over the last few months. For your reference, it also shows the 20th percentile value across all sites, separating slow and fast load times.”

 

graph1

 

As the graph above demonstrates the page load times had been increasing fairly steadily since February 2011 then peaked sharply in the middle of April to around 15 seconds. After the spike the page load continued to rise at around the same speed as it had been before reaching around 10.5 seconds by the 15th.

We put in a request to the client to look in to these speed problems having identified a few issues slowing the site down. Their server was upgraded on the 7th of June and an almost instantaneous effect was noticed. The page load time effectively halved to around 7 seconds and by the 10th of June the dwindling Google.co.uk keyword rankings were back up on the first and second SERP.

Now here’s what Google is saying about the site:

“On average, pages in your site take 5.1 SECONDS TO LOAD (updated on Jun 22, 2011). This is SLOWER THAN 76% OF SITES. These estimates are of LOW ACCURADY (fewer than 100 data points). The chart below shows how your site’s average page load time has changed over the last few months. For your reference, it also shows the 20th percentile value across all sites, separating slow and fast load times.”

 

graph2

 

It’s still slower than most sites, but it’s out of that bottom percentile where there is a danger of being penalised. Overall impressions and clicks are on the increase (as demonstrated below) as are keyword rankings.

 

graph3

 

After running rankings a couple of days later the impact was significant. All ten keywords were now ranking in the top 25 results – an increase of 100-200 positions almost overnight.

There are still improvements to page speed left to be implemented. Improvement such as combining Javascript/ CSS files and caching static content are still issues, however implementing these other improvements will not shave anywhere as much time off loading as the site does not have a particularly large page weight.

In conclusion, the reactionary response from the search engine once the server was upgraded was almost knee-jerk in its execution and goes to show that perhaps SEOs and webmasters ought to be paying more attention to usability metrics than some of the more common SEO tactics they are currently employing.

Greig Holbrook is the Managing Director of Oban, the international digital marketing specialist. If you are interested in more information about effective International SEO for your business, please contact us on +44 (0)1273 704 434 or email info@obanmultilingual.com