The findings come from the US Google Ranking Factors Rank Correlation 2013 study by search and social analytics company, Searchmetrics, which aims to identify the key factors that help web pages rank well today in Google searches. A follow up to the companys similar US Ranking Factors study conducted in 2012, the new study analyzes search results from Google.com for 10,000 popular-keywords and 300,000 websites appearing in the top 30 search results and picks out the issues that correlate[1] with a high Google ranking.
Five key findings from the study are:
- Google plus ones correlate more with Google rankings than other social signals
Social signals such as Facebook likes and shares, Twitter tweets and Google plus ones all correlate closely with Google rankings. But Searchmetrics analysis found that Google plus ones correlate more strongly (a correlation coefficient of +0.4) than any other social media signals. Facebook shares were the second most closely linked social signal with rankings – a correlation of +0.34. Tweets on Twitter and Pins on Pinterest have lower correlations of +0.28 and 0.29 respectively.
The pages in top positions in the search results stand out with a very high mass of social signals, explained Marcus Tober, Searchmetrics founder and CTO. Of course theres a lot of debate about whether social signals directly influence rankings or are just closely correlated with rankings – because highly ranked pages will get more traffic and so attract more shares, likes, plus ones etc.
At Searchmetrics weve done a number of studies that we plan to publish in coming months which indicate that social signals do in fact have an influence on rankings.
- Quality content is important for search rankings
Search experts believe the quality of content on web pages is an important ranking factor because Google is keen to raise the rankings of pages with good, useful information and Searchmetrics analysis seems to support this. Pages with more text and more images are positively correlated with Google rankings (correlation +.11 and +0.08 respectively) indicating that higher ranking pages have more text and more images.
If we assume that the existence of more text and images is an indicator of quality, then quality content is linked to higher rankings according to our study, said Tober. But we actually found that this relationship exists up to a limit of around 10,000 characters after this the correlation tended to decrease. So you cant just go on adding text in the hope it will continue to drive a more positive rankings boost.
- Keyword domains and keyword links have lost relevance
Search experts have long believed that sites that have a keyword in the domain name, such as carinsurance.com or paydayloan.com, tend to rank higher for relevant searches, but Searchmetricss study indicates this is now declining in importance.The study found that having a keyword in the domain name has only a low positive correlation (+0.03) down from 0.11 in 2012. While having a keyword in the URL or web page address has declined from a correlation of +0.04 in 2012 to an even lower 0.01 in 2013.
Only a year ago, factors such as keywords in domains and URLs were correlated more positively with better rankings, explained Tober. So our 2013 analysis indicates that Googles “Exact Match Domain” (EMD) update to tackle the issue – as well as the Penguin algorithm updates – seem to have done a good job.
- The number of backlinks remains immensely important
As in Searchmetrics 2012 study, the 2013 data shows that the number of backlinks (links to a web page from other sites) remains very closely connected to how the page ranks. In fact the number of backlinks for pages that rank in top positions has increased, even if the correlation has declined slightly from +0.36 in 2012 to +0.34 in 2013 (this slight decline in correlation is because all the pages ranking in the top 30 positions feature more backlinks on average than last year).“Pages with more links rank even better,” Tober summarizes. “However, its not just the quantity that matters, but more and more the quality. The more natural the link profile, the greater the prospect of a positive ranking.
To get the most benefit, a site needs to have a spread of links that looks natural – not artificially created by SEO experts.
This means a site should not simply have a large number of perfectly optimized links that include keywords it wants to rank for in the anchor text. In line with this, having a proportion of no follow links (links which do not convey ranking benefits) and links that contain neutral stopwords (such as in, and, to) in the anchor text as well as links that are just generic (like there, here, page) have increased in importance according Searchmetrics analysis.
A related finding is that backlinks which contain target keywords in the anchor text, a factor normally considered important by SEO experts, has decreased in importance with a correlation of +0.06 in 2013 (down from +0.10 in 2012).
The SEO visibility of the web page a backlink comes from is also very important now, said Tober. SEO visibility is our measure of how visible a site is within Google searches. So what we mean here is that if a site gets backlinks from highly visible, high ranking pages such as respected national news sites, then they seem to be more valuable having more of this type of link is closely correlated with high rankings.
- Brands are the exception to many rules
Brands seem to automatically rank higher in Google searches. And the analysis reveals that some features that correlate with high ranking pages, mostly on-page[2] factors, do not apply to pages from top brands in the same way as they do for non-brand web pages.
Brands rank high even if they fail to fulfill some of the most basic principles of SEO, said Tober. For example pages from brands are positioned in or near the top of search results even if they have less text on the page and they dont have keywords in the title.
Searchmetrics collated the data for its study in March and then again in June 2013 to ensure it takes account of the impact of Googles latest Penguin 2.0 algorithm update.
Our study provides an up to date snapshot of the factors that correlate with rankings on Google US, explained Tober.
He added, however, that while the study highlighted those factors that correlate closely with a high Google ranking, it is not possible to say that these factors definitively cause or influence the rankings:
We can say that, on average, more well-positioned pages exhibited these features and more poorly placed pages did not. But as Ive said before, only Google knows what influences rankings for sure – as only Google controls how the search algorithm actually behaves.
Tober also pointed out that SEOs analyzing the data should be careful about making hasty interpretations from the ranking factor coefficient numbers. Certain factors tended to have a low correlation because they are present on nearly all pages that appear in the top 30 search results (resulting in a low Spearman correlation coefficient).
These factors tended to be very basic on-page factors such as the existence of H1 headings, a keyword in the meta description and sitespeed. They are almost ever-present and should not be disregarded by SEO teams. You can find out more about this issue by reading our in-depth report about the 2013 ranking factors.
The detailed report which outlines the results from Searchmetrics US Ranking Factors – Rank Correlation 2013 study can be downloaded here: www.searchmetrics.com/en/services/ranking-factors-2013/download/. And an infographic highlighting the key findings can be found here: www.searchmetrics.com/en/services/ranking-factors-2013/download/.
[1] Correlations were calculated using Spearman’s rank correlation coefficient. A coefficient score of +1 implies a perfect positive correlation and a score of -1 implies a perfect negative correlation. In Searchmetrics analysis a high positive correlation coefficient occurs for a factor if higher ranking pages have that feature / or more of that feature, while lower ranking pages do not / or have less of that feature.
[2] On-page factors are those factors that the webmaster, publisher or their SEO team/agency can control directly, such as the text length, page titles, URL features and headings on web pages.