Do Social Links Affect SEO? A Correlation Study
Do “social links” influence search ranking? This is a big question people in our industry are asking right now. To my mind, it’s not a question we can answer definitively. This is because you’re asking about causation; “do social links cause high ranking?” Because there are so many other factors involved in ranking which also relate to social links (e.g. content quality) it’s almost impossible to prove there is a causation here. A more sensible question is to ask “what’s the correlation between social links and search ranking?”
Correlation can be measured, although it takes a lot of data collection and your data must be consistent. I recently undertook a correlation study to investigate this question, which allowed me to kick the tyres on the new SearchMetrics Essentials module.
For any given site, Essentials can provide you with:
- A consistent score for visibility in SERPs. Searchmetrics provides an SEO visibility score for over 60m domains on a weekly basis. The score is based on a weekly survey of the top 100 rankings for over 1.25m keywords. Ranking, search volume and a proprietary CTR algorithm are combined to produce each domain’s score.
- Total social links for any domain or page.
In addition to this data from SearchMetrics I also pulled link data for each site from Majestic. The sites I looked at were all UK websites from the B2C sector. So for every site I had:
- the total number of social links, split out by network
- the total number of links, referring domains and referring IPs
- a consistent measure of SEO visibility
Whilst the process was quite manual, SearchMetrics are updating so you can get this data for loads of sites in one standard report. That’s going to enable me to revisit in a few months and do a wider correlation study looking at the differences between b2b & b2c sites, start-ups etc.
After correlating all the data sets vs. seo visibility I could look at answering some questions; first up, is there a correlation between popularity in any particular social channel and seo visibility?
I wasn’t surprised to see a bit of a mixed bag really when it came to results. Because Google+ has just launched there weren’t a whole lot of +1s for many of the brands, and it seems like very few people are using LinkedIn shares, so it was almost not worth looking at those two channels, the dataset was so small.
The second obvious conclusion is that the number of referring domains and IP addresses are still way out in front when it comes to correlation with social visibility. No social factor could touch these link metrics, which we know influence ranking.
What I thought was interesting was that the raw number of links and most FB metrics had an equal correlation with search visibility. However, it’s still impossible to say whether FB links have as much influence on rankings as normal links. I came up with this model illustrating what might cause links / social links and SERP visibility to grow at similar rates for a page / site, and I think it also does a good job of showing just how complicated it is to attribute causation for increased SERP ranking.
This horribly drawn powerpoint model is designed to illustrate how a page can attract both social and normal links, as well as how relevancy is signalled. A page or site could get social links purely because it’s home to great content, or alternatively, some social links could be the result of the page/site being visible to users in SERPs and/or referring pages. But what causes that visibility in SERPs? Links? Yes, that’s been confirmed. Social links? Unfortunately, as the model illustrates, there are so many variables which may cause an uplift in normal or social links that it’s impossible to say conclusively.
So why do so many SEOs believe social data is, or will be, a part of the ranking algorithm? It all boils down to relevancy. When it comes to search, relevancy is a currency we can acquire or pass on. We do this through links; as we all know, links signal relevancy for a topic. Unfortunately, as a percentage of web users, the number of people who have access to this currency is limited. Putting a hyperlink on a webpage is bread and butter to people like us, but to your average man on the street, it isn’t something they could do easily. There are over 800 million FB users in the world, and only 140 million active domains. How many webmasters does that equate to? Not a whole lot when compared to the number of people who know how to click a “like” button. Relevancy is a currency stuck in the hands of an elite.
To many minds, it makes sense for search engines to bring social links into the algorithm, as search engines would place relevancy into the hands of many more users. Relevancy would suddenly become more democratic, and the algorithm would have many more points of reference with regard to relevancy. Results would become much harder to game.
The natural extension of this is to then examine how many usage data points are generated each day. You quickly start to see why search engines are developing ways of referencing this data with regard to attributing relevancy. In a similar vein to the above, it makes sense to posit that as Google’s machine learning and processing capabilities increase, user data will be used more frequently in formulating ranking. In this scenario, relevancy would be attributed to websites by all users, simply as a result of them getting online and using a browser, OS, toolbar or the search engine’s pages.
I hope in this piece I’ve been able to demonstrate:
- How difficult it is to work out whether or not social links are boosting domain’s SERP visibility
- New ways of gathering data for studying this topic
- Why there may be a link between links, social links and visibility
- Why SEOs think search engines will use social links and user data in future
With regard to actionable insights, I would say don’t believe the hype regarding social signals… yet. Clearly, the best use of your time is pursuing normal links from a variety of relevant domains. Once Google has elaborated further on how social data impacts its algorithm I would revisit this.
In the meantime, use social link data alongside normal link information to understand where good quality content is; on your site and your competitors. This information should be used to inform your content strategy. The 3-step model I use for brand content strategy goes like this
- Understand why your best content / your competitors best content is well like
- Work out how to make bad content better, by looking at examples of similar information in that outranks your own
- Pick new keyword areas to target which other brands haven’t focussed on. To target the SERPs for these terms, build content which provides the same information as current high-ranking sites, but leverage your brand’s credibility, reach and knowledge of the topic to improve upon what is currently available.