The new 2013 Searchmetrics’ study of 300,000 websites reported that social sharing behavior is even more strongly correlated with high page rank.

Top highlights of 2013:

  1. Keyword domains and keyword links have lost relevance
  2. Brands are the exception to many rules
  3. Social signals continue to correlate very well with better rankings
  4. Good content is always important: it comes to quality!
  5. The number of backlinks remains immensely important
  6. On-page technology remains one of the basics

The updated report is available in its entirety here: http://www.searchmetrics.com/en/services/ranking-factors-2013/

According to the report, sharing engaging content in social networks correlates with higher search rankings much more than traditional SEO techniques.  See the chart below.

SEO search ranking factors 2013Pretty much mimics the 2012 results, except that social signals are even stronger predictors of search performance.

Takeaways from this study for me include:

  • Sharing your content on Facebook and Twitter and getting people to engage with it is much more strongly correlated with SEO performance than all traditional SEO methods save one: the number of backlinks.  The quality of those backlinks wasn’t measured in this study, but Google is prioritizing quality.
  • As I hypothesized in my earlier version of this post, Google+ beats other social networks when it comes to SEO.
  • Many traditional SEO tactics aren’t as effective as social sharing.  This includes positioning Keywords correctly and making sure the right Keywords are in your URLs, Title and Description.

Search geeks, take heed: according to Eric Schmidt, social signals will become fundamental to ranking well – for human factor reasons that are hard to argue.  In his new book he says,

Within search results, information tied to verified online profiles will be ranked higher than content without such verification, which will result in most users naturally clicking on the top (verified) results. The true cost of remaining anonymous, then, might be irrelevance.

To learn more about the changing relationship between social sharing and SEO performance, check out SEOMoz’s Daily SEO Blog / Social Media section. Rand Fishkin’s blog team does an awesome job covering the really important topic.

Here are some recent posts about this subject:

About the Searchmetrics Study

The study analyzed Google search results for 10,000 keywords as well as billions of backlinks, Tweets, Google +1s and Facebook likes, shares and comments. The data was collected in early 2013. The correlations between different factors and the Google search results were calculated using Spearman’s rank correlation coefficient.

The corpus of content related to these keywords include:

  • 30,000 SERPs (search engine result pages)
  • 300,000 titles, descriptions, URLs, etc.
  • Approx. 150 GB of data
  • Approx. 600,000 AdSense blocks
  • Approx. 5,300,000,000 backlinks
  • Approx. 4,150,000,000 Facebook shares
  • Approx. 12,950,000,000 Facebook likes
  • Approx. 600,000,000 Facebook comments
  • Approx. 1,000,000,000 tweets
  • Approx. 330,000,000 Google plus ones
  • Approx. 14,500,000 Pins
Social Sharing Outranks Many Other SEO Techniques (Updated for 2013) by


9 Responses

  1. Very interesting but one thing in the graph troubles me — why is keyword in the page title tag (and all the items below that) NEGATIVE ranking factors? Did you filter out sites hit by Penguin and Panda in your sample?

  2. I just read the white paper, but, for the moment, I’m not buying that those on-page techniques (e.g., keyword in title) are negative ranking factors — it makes no sense nor do they offer a compelling explanation for this finding nor do they talk about how they’ve controlled for or factored out sites hit by Penguin. It seems possible we’re seeing the effect of a confounding or correlated variable: Penguin.

    I would would want to see them re-run their model but control for (or remove) sites that have been hit w/an on-page over-optimization penalty (Penguin). Many sites hit w/such a penalty would be keyword stuffed all over the place, e.g., the title tag, H1 tag, etc. (unless it was just bad links that got them into trouble) and the spearman correlation would pick that up, i.e., the negative correlation would have nothing to do with on-page techniques at all. Bad site that have lost SEO visibility is confounding variable they haven’t taken into account. That’s my guess about what’s going on with those negative ranking on-page SEO variables.

  3. I, too, would love to see this white paper and your analysis of it, Vernon, with more recent data – with all the Google Algorithm changes Panda, Penguin, EMD. Thanks.

  4. Janet and Christopher – the report is available in its entirety here: (pay with a tweet to access it)

    http://www.searchmetrics.com/en/white-paper/google-ranking-factors-us-2012/

    Cheers!

  5. Excellent information, thank you. I’ve been seeing this for a while now and thought it was always obvious Google+ was going to take over backlinks eventually. Someone recommending or “+1’ing” a piece of content is like a 100% verified backlink, as Google can easily pluck out fake Google+ accounts.

  6. This is progress.

    I don’t agree with keywords in title giving negative ranking factors tough. It defeats the point of having a fully indexible www.

  7. “Christopher, This study was published in June, so the analysis period was probably prior to Penguin’s launch in April 2012.”

    Penguin was first rolled out on April 24, 2012. You should double check to make sure no data was used after April 24th 2012? If site data was used, you’ve got a possible confound in your data. Also, before 4/24/12, many sites were getting hit hand with manual penalties, i.e, unnatural link warnings and those sites took a tumble in traffic. Sites that built unnatural links were also, probably, more likely to do on-page over-optimization so more of those sites than non-penalized sites may have had a keyword in the title, in the H1 tag.

    In short, echoing Janet below, I too would have more confidence in the study if it had explicitly controlled for — better yet, removed — all sites that have been hit with Penguin or Panda (it may not be possible to identify sites hit with manual action due to unnatural links however). Panda hit sites are also probably more likely to be over-optimized.

    My guess is your basic finding, social sharing behavior is strongly correlated with high page rank, will survive, but if the negative correlations for those on-page factors disappeared, or got significantly smaller, that would suggest this study’s data is confounded with sites hit with an over-optimization penalties — see en.wikipedia.org/wiki/Confounding for more about uncontrolled confounding variables.

  8. Thanks for your insight, great points.