There is a lot of buzz and fuzz regarding the influence of shares, comments, likes, etc. on SEO in general. Yet, the question that is on everyone’s lips is actually: “Are social signals a ranking factor?” There have been lots of discussions around this topic, both pros and cons; even some studies were conducted on it. Yet, no clear light has been shed on the matter. If social signals affect SEO, how does it happens?
And, as we think that research is one of the most exciting and, most of all, rewarding of occupations, we’ve conducted an in-depth investigation in order to find out exactly this. What we wanted to see was whether there is any good reason to believe social networking sites have any relation to page ranking beyond anecdotal evidence. In order to study this relation, we needed quite a bit of data, therefore I warn you that you might need to arm yourself with a lot of time and coffee before reading our study.
TL;DR – This is quite a large study. If you don’t have time to read it all now, you can browse through the main take-aways.
- The Methodological Approach – How the Research Was Done
- Strong Presence on Social Networks Is Correlated with Better Rankings
- Higher Rankings are Correlated with Facebook, Google +, LinkedIn & Pinterest High Shares Altogether
- Top 4 Ranking Positions Have Significantly More Facebook Activity
- The Average Google+ Shares for the 1st Rank is Significantly Higher
- No Direct Correlation Can Be Made Between LinkedIn Activity and Ranks
- Pinterest High Shares Don’t Correlate at All with High Rankings
- Sharing Activity Correlated With Rankings and the Content’s Length
- Content Between 1 – 50 Words Is Correlated with High Facebook Activity and 1st Ranks
- There Is No Correlation Between FB Activity and Ranks when We Look at Long Content
- Micro Content that Ranks 1st Is Correlated with High G+ Shares
- Slightly Linearity Between G+ Shares and Ranks for Long Content
- LinkedIn High Shares Correlate with Micro Content Ranking 1st
- Pinterest Shares Do Not Correlate With Ranks When Looking at Micro Content
- Linear Correlation Between Pinterest shares and Ranks When Looking at Long Content
- Some More Methodological Clarification
- Other Relevant Studies & Google’s Position
The Methodological Approach- How the Research Was Done
We wouldn’t want to bore you with a lot of technicalities related to the research, however, allow us to present you the study’s main points:
- The current study is based on social signals coming from Facebook, Google+, Pinterest and LinkedIn. We couldn’t include Twitter into this analysis due to their decision of deactivating share counts.
- We took into consideration all social signals from approximately 300k pieces of content, coming from ~34k randomly chosen keywords that rank on Google on positions 1 to 10. All the research data were gathered over the months of May and June 2016.
- The current study shows a correlation between social signals and the search engine Google. Yet, correlation does not mean causation.
One methodological clarification we’d like to mention is that the mean values were calculated for all entries which had at least one like, share or comment (therefore greater than 0), but less than 100.000. The upper limit was set for both theoretical considerations (social signals presence above that level tends to be rarely organic) as well as pragmatic ones (given the scale, even a small number of values over that threshold can influence a rank’s score and make it unrepresentative).
Strong Presence on Social Networks Is Correlated with Better Rankings
On average, social presences (which includes likes, shares and comments on Facebook, plus shares on Google+, LinkedIn and Pinterest) are negatively associated with site rank, and the relationship is close to linear (and perfectly linear for the first 5 ranks). This means that, in general, the smaller the rank number is (so, the higher up the website), the higher the chances are that the average presence on social network is larger.
Higher Rankings are Correlated with Facebook, Google +, LinkedIn & Pinterest High Shares Altogether
This holds true for almost all of the individual social channels as well, though the linearity of the relationship varies. Facebook (overall activity, including likes, shares and comments) and Google+ are the closest to a perfectly linear relationship, each with 2 “deviations” from the expected values. LinkedIn’s relation to site rank is decidedly less linear, although the overall trend still holds true. The one site that stands out is Pinterest, because there is no linearity whatsoever, and also because it is the only social platform where the highest number of shares is not associated with the first rank.
Top 4 Ranking Positions Have Significantly More Facebook Activity
Even with Facebook, things are slightly more complicated. For Likes, the linearity is broken from the second rank, but except for that it holds true for the first 5 ranks and the top 3 all have significantly higher numbers than the rest.
More Shares are associated with higher ranks (so, then, lower rank numbers) in a clear fashion for the first 7 ranks and, again, the first 3 ranks are significantly higher than the rest (with the first being significantly higher than the next two).
Comments maintain the linearity for the first 6 ranks and the first rank has significantly higher numbers than all that follow. Overall, while no causality can be inferred, more presence on Facebook is clearly associated with a higher rank on the search list.
The Average Google+ Shares for the 1st Rank is Significantly Higher
With Google+, the linearity is still there, overall, but more so in brackets. There is a clear streak from the 1st to the 4th rank, than another from the 5th to the 8th. Furthermore, the mean for the first rank is significantly higher than the other values, as is the difference between the mean values for the first two ranks, compared to differences between any other two ranks.
No Direct Correlation Can Be Made Between LinkedIn Activity and Ranks
For LinkedIn there isn’t much of a linearity, save for, maybe, the one that can be seen for ranks 6 through 8 (but to draw any conclusion about all ten ranks would be a stretch). What can still be said with certainty is that the mean for the first rank is higher than for any other rank. Still, it’s difficult to claim anything about associations beyond that, since the second highest mean value corresponds to the 10th rank, while the lowest mean corresponds to the 8th rank.
Pinterest High Shares Don’t Correlate at All with High Rankings
Pinterest is the one social network that stands out for two reasons. The first is that there is clearly no linearity whatsoever (at least not in the direction of the same relation that we’ve seen for Facebook and Google+). The second reason is that this time, the highest mean value of shares isn’t even associated with the first rank. The two highest values are associated with ranks 8 and 7, respectively. In fact, the mean for the 1st rank turns out to be the 9th highest (and ironically, the only mean that’s smaller is the one for the 9th rank). This is not to say that you shouldn’t try to be getting as many shares on Pinterest as possible, but rather that they will most likely not be a good indicator of overall search results ranking prowess.
There is a strong correlation between ranks and social activity in general, yet, each social network needs to be analyzed in particular in order to draw the right conclusions.
Sharing Activity Correlated With Rankings and the Content’s Length
In an earlier study conducted in the cognitive labs we figured out that shorter articles are usually correlated with higher rankings. But is it a connection within a post’s length, its number of shares and rankings? (consider rephrasing) With this precious data at our disposal, we’ve decided to see whether there is a connection between the content’s length, the number of shares and rankings.
We’ve classified all >300k posts into categories that went from 1 to 10k words. We are going to stress on two categories that we found more relevant.
- Content between 1 – 50 words. We’re basically looking at micro-content here. We chose to take this category into discussion as, usually, the first thing that comes to mind when we talk about social media is short content.
- Content between 1001-5000 words. We chose this category as this length of content is the “winner” in terms of rankings, as you can see in the screenshot below.
Content Between 1 – 50 Words Is Correlated with High Facebook Activity and 1st Ranks
As predicted, when it comes to micro-content, we can see a huge difference between the first position and all the others. Even if we sum up all the other nine positions’ number of shares and we still don’t get the high number of likes, shares & comments that the first rank holds.
There Is No Correlation Between Facebook Activity and Ranks when We Look at Long Content
When it comes to posts that are usually successful judging by the number of words, we cannot say that the same success replicates when it comes to the number of Facebook activities for the first ranks. We are talking here about the number of Facebook shares, likes, and comments altogether. Yet, a direct correlation between Facebook activity and ranks for content that has between 1001 and 5000 words cannot be made.
Micro Content that Ranks 1st is Correlated with High G+ Shares
As in the case of Facebook, micro-content seems to make a real breakthrough. Yet, although we can see a high correlation between the first position and the number of Google+ shares, same linearity does not apply for the rest of the positions. For instance, there are almost the same number of shares for the second position as there are for the 7th. However, a strong correlation can be made between micro content, high Google + shares and first organic search ranking position.
Slightly Linearity between G+ Shares and Ranks for Long Content
When it comes to content between 1001-5000 words we can see a sort of a linearity which, although not a consistent one, is still better than in the case of Facebook. Except the third position, all first six positions seem to be following a sort of consistency when talking about the relation between G+ shares and ranks.
LinkedIn High Shares Correlate with Micro Content Ranking 1st
If we couldn’t determine a strong correlation between ranks and LinkedIn activity when it comes to the total sample, when looking at content with 1-50 words it looks like we can definitely say that there is a correlation between LinkedIn shares, micro-content and ranking in search on the first position. Then again, just like in the previous cases, there is no direct correlation when it comes to the other ranking positions; yet, the relation with the first position is so strong that it might raise some interest.
For 1001 – 5000 words content, LinkedIn seems to correlate with ranks just like it usually does…or, better said, doesn’t. Just by taking a look at the chart below we can figure out that there is not a strong correlation between Linkedin Shares, medium length content and ranks.
Pinterest Shares Do Not Correlate With Ranks When Looking at Micro Content
Pinterest continues to have the same peculiar behavior no matter of the sample of content and ranks we are looking at. Indeed, micro content ranking first has by far the most Pinterest share, yet, we cannot talk of any sort of linearity as all the other positions don’t seem to express that. Just by taking a look at the chart below we realize that “correlation” or “linearity” are not the most suitable words to be used in this context.
Linear Correlation Between Pinterest Shares and Ranks when Looking at Long Content
As mentioned earlier, Pinterest seem to be special when it comes to posts with 1001 – 5000 words, only that this time, in the sense that it’s the social network that has, probably, the most clear linearity between the number of shares and ranks when it comes to medium content.
Some More Methodological Clarification
Please indulge us while explaining why we chose to look only at the social signals coming from top 10 organic search results. In search engine pages, this usually means that we only looked at the first page of a web search. We chose to do so because, to be honest, if you’re not in the top 10 there are really small chances for a user to find your page by performing a search. There are several studies on this and results may vary, but there’s a world of difference between being on the first and being on the second page. As you can see below, even by the most conservative of estimates, the difference between being on the first page and being on the second is quite staggering.
Chitika study: https://chitika.com/google-positioning-value
In terms of pages, the Optify study claims the first page of results rakes in about 89.69% of the results, while the Chitika one puts the number at 91.5%. The Moz study, by comparison, puts that number at 52.40%, a much lower figure. It’s also the more recent study, so there’s a possibility it is slightly more accurate.
Other Relevant Studies & Google’s Position
As we were mentioning before, a lot of discussions and even studies were conducted on the importance (or lack thereof) of social signals in rankings.
While some of them, like the Moz study, claim that there is strong reason to believe Google doesn’t use social share counts directly in its algorithm, there are other positions on this, like Neil Patels’, that highlight the fact that there might be a strong connection between the two (just like we can see in the screenshot below taken from the Quick Sprout’s gifographic).
There are several opinions about the role of social signals in organic SEO rankings. Of course, even Google, through Matt Cutts (currently on an extended hiatus from his job as head of Google’s web spam team) had something to say about this. Long story short, what the search engine communicated in 2014 was that Google treats Facebook and Twitter posts like any other web pages for search, but NOT as a ranking factor. And why would they do so? Because, as Google says, they won’t use a signal to influence its search rankings unless they have high confidence in the meaning of that signal.
According to Cutts, one should be active on social networks for many good reasons, yet, ranking high wouldn’t be one of those. John Mueller, Webmaster Trends Analyst at Google also insisted that that there is no direct ranking signal in Google’s ranking algorithm.
We cannot agree more that one should be present on social media not for rankings but for building up their brand and driving qualified traffic. Yet, the present study made us wonder if indeed Google is doing what it’s preaching and whether Google’s engineers don’t “monetize” social signals.
It’s the shares that lead to a better position, or sites with a higher position naturally get more shares?
Now that all the numbers are in, it’s important to set one thing straight. It’s the old scientific motto that correlation does not imply causation. To be fair, we’ve never claimed we have been trying to prove (or disprove) that a higher presence on social networking sites will lead to a higher place in search engine rankings. We’re merely observing that there is some relation between the two, though the exact nature is probably more complicated than that and might not even be the same for all the social networks.
Even the tests that we did on the data aim merely at establishing the strength of a correlation, not the directionality of a causal link. So we know there is some relation between shares and position in the rankings list, we’re just not sure whether it’s the shares that lead to a better position, or sites with a higher position naturally get more shares.
That being said, it’s probably also worth mentioning another recent adage, coming from the creator of the XKCD web comic, which states that “correlation does not imply causation, but it does waggle its eyebrows suggestively and gesture furtively while mouthing ‘look over there.’“ Which is to say that just because we cannot definitely state that more shares on a social network will lead to a higher position in the rankings list, it doesn’t mean it can’t happen. We just have to take other things into account as well.
Who did this research
- Razvan Gavrilas Researched & Audited the Analysis
- Cornelia Cozmiuc Researched & Wrote the Paper
- Ionut Astratiei Performed the Crawlings
and Data Validations