Everyone in search is by now aware that certain social signals are well-correlated with rankings.
In each major study published on the subject, the authors point to how correlation does not imply causation (see, for example SEOmoz and Searchmetrics). Dr. Pete even wrote a whole post on the subject.
I wanted to see if it was actually plausible for these correlations to arise without social signals being a direct ranking factor. I built some Excel models to test this out and see if I could build a model that achieved the observed correlations without assuming social signals as a ranking factor.
The punchline: it's possible there is no causation
I have a suspicion that this could be the most misinterpreted post I have ever written, so I thought I'd start with a prominent "Cliff notes" to be explicitly clear about what I am saying and more importantly what I am not saying.
I am saying
You can tweet any of the following without misrepresenting me:
- Social signals *may* be correlated with better rankings but not cause them [tweet this]
- Facebook Likes and rankings could achieve high correlation without Likes being a ranking factor [tweet this]
I am not saying
If you tweet any of the following attributed to me, I will write "does not follow instructions" on your forehead in magic marker:
- Likes don't matter [where did I say that?] [tweet this]
- Likes aren't a ranking factor [I don't show any evidence either way] [tweet this]
- Links are dead [what?] [tweet this]
- Correlation studies are a bad idea [I agree with Rand that we could actually use more studies] [tweet this]
What is this based on?
I have built a simplified Excel model of how pages accrue Likes over time. With no assumption of them being a ranking factor, I nevertheless demonstrate that we could see a strong correlation between Likes and ranking position.
Why focus on Likes?
The modelling works equally well with any of the social signals. I simply chose Likes to make the example more concrete - you could build the exact some correlation model with Tweets, Facebook Shares, Google +1s, or any other signal where accruing more social shares makes it even more likely that you will accrue more in the future.
Starting at the beginning
Every time we see a correlation study, I see evidence that some people haven't completely taken on board the correlation/causation subtleties. This is unsurprising - the mathematics behind the calculations in these study is typically undergraduate level (with some of the advanced analysis verging on graduate level) - most people's intuition lets them down horribly when confronted with probability and statistics. (Don't believe me? Check out the Monty Hall Problem).
So let's start from the beginning:
What are these studies looking for?
When we say correlation in this context, you can imagine that what we are looking for is similarity. We are looking for evidence that two things happen together (and don't happen together).
In the context of these studies, we are typically looking to see if "ranking well" happens together with "strong social signals."
Now - the mathematical part comes in when we try to define "happens together with" properly. The human brain is a remarkably powerful pattern matching device. For example - how many sportsmen and women have a pre-game routine involving a specific pair of lucky socks because of a sequence of events something like:
- Wore a new pair of socks today. Kicked ass.
- Wore the same pair of socks as last week. Kicked ass.
- New socks in the wash. Grabbed a different pair. Got whupped.
- Socks successfully cleaned and dried. Kicked ass again.
Pretty compelling evidence for those socks, huh?
From that point onwards, the athlete refuses to surrender the lucky socks. Any future losses are attributed to other factors ("I did everything I could - I even wore my lucky socks").
Michael Jordan apparently started wearing longer shorts to cover his UNC "lucky shorts"
But let's look at this a little more closely and skeptically. Are there any other explanations for this sequence of events? Imagine that the athlete in question is good - winning roughly 75% of his or her games on average. Imagine also that the socks are, in fact, not magic and that they have no impact on the result (shocking, I know). The odds that the single loss of a set of 4 games will coincide with a single wear of a different pair of socks is then: 0.75 x 0.75 x 0.25 x 0.75 = 0.11
In other words, roughly one in ten pairs of socks would randomly look this lucky.
Given all this evidence, most of us would probably chalk it up to chance (but keep wearing our lucky socks just in case).
Add to this the fact that we can't help but be always on the lookout for these patterns (it's just how our brains are wired) and it's unsurprising that there is always some pattern to be seen somewhere.
Given all of this, we apply pretty high standards of proof before stating that there is correlation [i.e. that two things tend to happen (or not) together]. This is measured with a "confidence" which is similar to the layman's definition but is measured in probabilities. We express our confidence in terms of "the probability that we would see a correlation at least this strong even if there were no underlying correlation." Statisticians typically talk in 95% or 99% confidence ranges (though note that a 95% confidence interval is still wrong one time in 20).
The ranking factor studies undertaken by SEOmoz and others have shown a non-zero correlation with high confidence. In other words, there is a correlation between certain social signals and higher rankings. I don't think anyone is seriously disputing that at this point.
Correlation is not causation
This tricky phrase gets wheeled out with every study. What does it mean?
It means that the mathematical techniques we have applied to be confident that there is a relationship between these two variables says nothing about whether one causes the other.
It's easy to think of correlations that are not causative. More ice creams are sold in months when more sun lotion is sold. Sun lotion sales don't cause ice cream sales and ice cream sales don't cause sun lotion sales. Both are caused by sunny days.
While measuring correlation is straightforward based solely on raw data, this is generally not sufficient to judge causation. This is especially true where neither variable is in your control (such as the sunshine example above). Measuring or understanding causation is a topic for another day.
The important thing to note is that the size of the correlation or the degree of confidence in the correlation have no bearing at all on the likelihood of a causal relationship.
This is one of the common misconceptions with the interaction between social signals and rankings - when people say things like:
"But the correlation is too high - social signals must be a ranking factor"
I'm afraid my response is
"I'm sorry to inform you that you have been taken in by unsupportable mathematics designed to prey on the gullible and the lonely."
Sorry for the Sheldon moment there
Seeking alternative explanations
I believe in a healthy skepticism when presented with bold evidence. I can see lots of arguments why search engines could view social signals as ranking factors (though at least in the case of tweets, I've long supported an algorithmic discounting of nofollow). For all the reasons outlined above, however, I'm not convinced we have seen real evidence that this is in fact what is happening.
Assuming we take the correlation studies at face value, there are three possible explanations:
1. Social signals are a ranking factor (and apparently a strong one at that)
This appears to be the hypothesis of Searchmetrics:
These findings come from a study by search and social analytics company Searchmetrics aimed at identifying the key factors that help web pages rank well in Google searches
From Searchmetrics (emphasis mine).
2. The causation goes the other way - ranking well results in better social signals
Although it's hard to know how strong this effect could be, it's easy to believe there is some kind of effect here. Just think about your searching/liking behaviour:
Carry out a search:
![Search for [excel for seo]](https://d1avok0lzls2w.cloudfront.net/img_uploads/excel-search.png)
Click on a link:
Recommend the page:
I only used this example because I know how disappointed Mike was when we had to move this page - while the redirect carried across much of the link equity, it reset the social signals - this content has been tweeted and Liked thousands of times. Sorry, Mike.
3. There is a hidden causal variable (some kind of "page quality" signal?) that causes better rankings and increased social signals
The research Dan Zarrella published here last week on the relationship between social signals and links indicates that this is a plausible explanation - since we see that there is a fairly strong relationship between the two. The challenge with this approach is that if we believe social signals' correlation with rankings comes entirely from their correlation with a real ranking factor, it's surprising that we often see a stronger correlation between rankings and social signals than with any other single factor:
Can the alternatives account for the observations?
Whenever this has come up in conversation, I've had people express doubts that #2 or #3 could be strong enough effects to give the results we see.
My intuition said that #2 could be. Mainly based off the fact that any effect that is there will compound over time under an assumption that "Likes beget Likes" which seems reasonable given the way that Facebook edgerank and visibility work. If we have compounding growth to magnify small effects, then over time we could see remarkable correlation appear from relatively small effects.
So I decided to see if I could build a plausible model of #2.
Imperfect models
What do I mean by a plausible model?
I mean that I'm going to simplify a whole raft of stuff from the real world (I'm going to think about a single SERP, for example, and I'm going to think only in time units of months). I'm going to attempt not to have these oversimplifications bias the answer in my favour. My default position (known as the "null hypothesis" in statistics and probability) is that these effects are not strong enough. I'm going to construct a model that biases towards that being true and see if I can still produce a strong enough effect.
Hacking the Excel
Editor's note: The Excel model discussed below is no longer available.
I built this model in Excel [warning: macros]. It's very hacky - just designed to find an answer rather than to be a robust model. It takes a set of simple assumptions (none of which include a causal link from Likes to rankings) - you can see these on the "Input" sheet - and you can substitute your own values if you would like to see the impact these have:
- Top ranking pages get 400 visits / month from search (the model over-simplifies to think about a single keyword/SERP getting 1,000 searches a month - this is a proxy for all organic traffic to the page)
- Organic traffic drops off through the ranking positions according to an averaged traffic distribution
- Each website is labelled as doing "Facebook marketing" (whatever that entails exactly) with a 30% probability. Facebook marketing doubles the rate of "random" Like acquisition. <geeky details>Sites not doing FB marketing in the model accrue "random" Likes according to a Poisson distribution with a mean of 10</geeky details>
- Likes --> more Likes at a rate of 3% (i.e. for every 100 Likes a page has, it'll get 3 more in the next month)
- Traffic --> Likes at a rate of 1%
This is what the Poisson distributions look like for the geeks in the audience:
.png)
It creates a really simple time series of Likes for each page in each month. The model runs for 36 time periods. Each refresh of Excel runs a new scenario and results in a single Spearman rank correlation at the end of the time period. Spearman rank correlation is the same measurement tool used in the SEOmoz and Searchmetrics studies.
This is what the growth of Likes looks like for a single example run (note that the lines are not ordered by ranking position despite the fact that the ordering is correlated with ranking across many runs):
.png)
I then ran the same model a hundred independent times to get a fair assessment of the correlation we could expect as a result of the simple assumptions above. (There is an embedded macro that does this for you if you would like to reproduce it - I'm not a macro expert - there's no doubt loads wrong with this):
I got a correlation of 0.44
This is actually higher than the correlation found in most studies I'm aware of and was based off my first pass of "finger in the air" assumptions. It's easy to tweak the assumptions to get way higher. I'd be interested in a discussion about realistic assumptions and/or flaws in the methodology.
Since there is definitively no causation in my model, unless someone can find a flaw in my method (a very real possibility - I'm a little rusty at this), I'm going to declare that it absolutely is possible for the factors we described above to be strong enough to result in the measured correlation without Likes causing better rankings directly. (Remember - you could build this exact same model applied to any of the social signals so this applies equally well to Tweets, Facebook Shares, Google +1s, etc.)
I'd love to hear if you think I've missed something or got something wrong in building my model.
Finally, the healthy skepticism needed around social signals. Great post.
I liked your illustration about the ice-cream and the suntan lotion. The key is SUMMER. In the summer, people seek both these things.
With ranking, the key is QUALITY. If the page is of high quality, people tend to both link to it and share it.
Sometimes it really is that simple.
Correlation is not Causation, Correlation is not Causation, Correlation is not Causation...maybe if people repeats it as a mantra, it will finally understand that correlation is not causation :)
So, you are saying correlation is causation? Right?
Here's a hypothesis for you that I think might hold some water. We know Google tracks your web history. We know they see your every move in Chrome. We know they tracked you in Safari. In Google Analytics you can see a referrer report and see where your traffic comes from.
I have a (thus far) un-testable hypothesis that your traffic is a major player in your ranking. We know links help, but you had better believe Google isn't just looking at links but who clicks those links if anyone. We know Facebook Shares seem to help, but can Google crawl all of the content on every Facebook page and profile? No, but they can see any traffic a site gets that was referred from Facebook.
I'm telling you, if we could pool a lot of analytics data together you could look at this. Look at how many referral sources there are. Look at the authority and trust of those referral sources. Look at what percentage of a site's traffic comes from search as opposed to referrals or direct. I have no doubt these things are playing a role somehow, I just don't have any kind of test I could run to prove it.
Good point Dan. I would like to add my 2 cents. More likes indicates more unique traffic and Google does take that in ranking consideration.
Dan and Sangeeta - I like the additional impact factor thinking. I will layer my 2 cents on top of each of yours having to do with referral and engagement metrics collected from a browser etc...
Assume that 6 shares = 3 verbal conversation about said content and 3 verbal conversations (OMG you have to check out article XYZ!!!!) propel 1 very specific search query likely containing a defining (KWD e.g. XYZ) for that content. When this user hits the content from search he/she will be highly engaged so browser/algo is now collecting multiple positive engagement metrics on the article. So again we have another factor (increased social - that promotes verbal social - increases search engagement). That could likely be correlated alone but is more likely one of many signals driven by the search and social ecosystem.
Thanks for sharing, Will! There's a lot in here to take away beyond just your research. You've helped some (me included) actually do this type of research.
How would you go about testing the half-life of a Like (or any social signal)? Is it probable that social signals only impact the ranking algorithm if the query deserves freshness signal is triggered?
Thanks again for the research!
This an excellent question. And, in addition to the freshness of a social signal, what about the quality of the person who cites it?
I have a feeling that not all tweets, likes, pluses, etc are the same, and depending on the individual and their level of influence, one citation could mean something completely different than another. Just like the author attribute conversations lately, (specifically how it might be a quality signal of content if the piece has comments from authoritative people) I wonder if social signals will be factored in terms of quality, not quantity.
I just got transported back to my undergrad stats classes. My thought is completely anecdotal, but I feel like #2 ("The causation goes the other way - ranking well results in better social signals) is an accurate description of what is happening
My thought is that the way social media is weaved into everyone's daily life, the shares are a result of a popular page that probably has a whole host of more probable ranking factors going on. Pages with a highly engaged social following most likely are gaining more links as a result of highly shared content. Also, interaction with the page will be heightened resulting in more visits and probably lower bounce rates.
Google is going to continue to claim that these social signals are not a ranking factor for the time being, so we will just have to continue to check out awesome posts like this and make our own assumptions. I think we can definitely say that creating a highly engaged community and accruing social signals is a good thing for marketing websites though
Great article, but doesn't it just come down to Google not owning Facebook and Twitter? They don't have privileged reliable access to those signals, except via a crawl, which is a bit flaky on datapoints like counts and tying likes back to influential profiles. Expect Google+, and circles in particular, to be the social signals that matter. Danny Sullivan for instance is in ~1.5 million circles—inevitably, most of them using "SEO" in the circle name. A +1 from him is going to be qualitatively different than a +1 from an average joe. It's just a matter of Google building up critical mass with circle membership and +1 votes. Could be years, but it will eventually "cross-over" the link graph and PageRank as a quality relevancy signal.
We can all predict that, but it seems a bit flaky as a business strategy for Google.
1) Are they willing to invest continually and without results in Google+ for many years to reach critical mass? They haven't shown that in the past.
2) With Bing's integration with Facebook isn't not a bit dangerous to stay put? I feel like they have the crawling capacities to figure it out quite well. Bing is not on the map per say, but a Facebook acquisition could change that in a hurry.
3) They can't really depend on ONE social network to bring enough social signals to the table when all sites will have social component in the near future, plus the incoming embedding-palooza. They *have* to develop this capacity in my humble opinion, I don't think it's optional as the signal structure of the web evolves no matter what.
4) I think they are banking on a big change in the input interfaces to replace the current signals (Google Glasses, Siri-Like VI assistants) more than hoping social signals will be the next big thing.
Yes yes yes! Every time I see a blog post about social signals correlating with links or rankings I scream at the monitor.
My assumption a lot of the time is that Z (the outside factor) is the sheer potential for popularity of the topic and the site/page itself. For example, this year in America all anyone can talk about concerning the weather is that this year is unseasonably hot/stormy and they think it's due to climate change. This common concern of people, I can almost certainly posit, has caused pages on sites about the weather and climate change to increase in links, rankings, AND likes/tweets.
We as technical-minded marketers tend to forget about the sociological basis of phenomena, and always try to chalk it up to the numbers. I might write a blog post about this issue...
Yea... but.
I've seen this at least a few times:
An article with X inbound links ranks page 2 on SERP.
Suddenly XY new tweets about this article appear and article emerges to 1st spot of the page 1 SERP.
A few days later article goes a few spots down. Number of inbound links is still X.
I didn't say they aren't impacting rankings - just that there are other possible explanations for the correlation studies.
However, the plural of anecdote isn't data.
I'd love to see a causation study that looks at this stuff on a bigger scale.
Indeed you didn't, my mistake :)
But, I'm just saying, I don't think that the source of our high ranking correlation of social signals is the links they get due to new traffic, since I've seen it a lot of times that amount of links remains the same and rankings go up with new tweets. It might be some other factor that isn't inbound links, but I have no idea what it could be. For now, I think I'm sticking with the theory that social signals can at least temporarily positively influence ranking position.
Freshness algorithm?
Absolutely agree with you galileo. Also There are nothing about new content and so-called "newbie bonus". For new unique content you *may* get better positions in SERPs even without any social pings, but later you *may* get down and down...
Good article though. Have a nice day!
I'll agree with that - but it's also from anecdotal evidence. I ran a blog that had very few inbound links to it. I produced a blog post with a "tweet to download files" on it which received 600+ tweets / RTs. The blog went from a PR 0 to a PR 3 literally overnight.
Now, I know that this could have coincidentally coincided with the toolbar update so I can't know for definite. However, I feel (what a horrible phrase) it did contribute quite heavily.
Would be good to see some proper studies on this.
It's more likely that Google happened to update toolbar PR the night you added your post.
Isn't this is a side issue though? PageRank is not the same as rankings so this example doesn't provide any material to inform this debate.
All though a tweet is a social signal, it is possible that Google decided to count the link in the tweet as a link without nofollow. I agree with Will that it is highly likely Google chooses when to follow/nofollow some links completely disregarding rel="nofollow" that Twitter and other Web platforms automatically add into the link.
I've also wondered if Google uses social signals as a way to "predict" the link graph. For example, a site gets a ton of social activity so Google expects links to follow at a later time (say within 3 days). If the expected links materialize the site maintains its rankings, if they don't the site loses ground.
No study or proof though, just a hypothesis ;)
Perhaps when something gets 600 tweets is would naturally get some links as well... When I find a good article that I can use for one of my articles I would "naturally" link to it.
Exactly. The problem is that people look at the tweets themselves and think they create correlation.
The reality is that many people, especially in tech industries, have their tweets automatically displayed on their website, and they are almost always dofollow. Hey presto, you get lots of links to your page from people's websites. Some time later, depending on how often they tweet and how many tweets they display, the link disappears off their site. You lose a lot of links, your rankings tank again, you assume it's a 'freshness' factor from tweets.
I would like to see someone prove/disprove my theory, but I don't think ahrefs's new/lost link tool is accurate enough for the job.
Jenni,
Yea, that could be it. But if I was an engineer at Google, I'd implement some sort of algo that knows the most common ways of embedding Twitter stream and count those links as tweets. Since they have a lot of smart people over there at Google, chances are someone already thought of it over there.
look at socials signals the same way you look at google hottrends. They are not a true indicator of value. The real stuff does not get xnos of tweets , it get x number of sources, and referrals.
Atleast that how i see it. Another point you should consider is google relationship with companies that "own" social platforms.
Here's one small argument against the reverse casual argument, where higher rankings lead to more sharing.
I find it interesting that the correlation between rankings and social activity is strongest for Facebook "shares," rather than Facebook "likes." I don't have any data to back it up, but it seems to me that people are far more prone to "like" than to "share" something that they come across on the web.
If higher rankings were leading to increased social activity, I would expect a stronger correlation with "likes" as opposed to "shares," the reverse of what we're actually seeing.
While that's certainly not a rigorous argument, it gives me reason enough to lean toward social activity as a ranking factor. At least in the case of Facebook shares.
One way to possibly get more information about the influence of social signals and maybe find a real causation is to have a "ranking over time" data stream with controlled input of social signals and links. It's still wouldn't be perfect, because others factors could cloud the study, but I would be curious about the results. You may not see anything interesting if you do only 4-5 tests either, you may want to run 80-100 tests with different chronologies and in extremely well controlled conditions.
What would also be interesting in that study is to see the chronology of events.
- Is social an enabler of links over time? (A social blip, followed by a link blip, followed by a ranking blip on a graph)
- Is social a weak signal by itself? (Double blip in the ranking pattern)
- Does ranking drive social? (Does social signals come after rankings? Ranking blip, followed by social blip)
etc.
Really interesting ideas and well executed Will. I think you did a great job in this post illustrating the possibility that there could be a reverse correlation between rankings and shares.
I'm wondering if you did any sensitivity analysis with your model parameters to see how they impact your conclusion? I'm guessing it might be pretty sensitive based on your comment at the end "It's easy to tweak the assumptions to get way higher."
Thanks Matt.
I guess a sensitivity analysis could show some pretty significant non-linearities I imagine. I don't know much about the best way to investigate that. Where should I start reading?
Hi Will,
I'd start by making a few plots that show the final overall correlation vs a model parameter. It looks like you have about 5 or 6 main parameters in your Excel table that will impact the output. Pick one, let's say "Conversion rate of traffic to likes" that is set to 1% now. You can set it to each of the values 0, 0.5, 1, 1.5 2, 5, and 10%, while keeping all the other parameters fixed and compute the Spearman correlation for each. Then make a plot of correlation vs the model parameter. If the model is very sensitive to this parameter, then the resulting correlation will change a lot. If it isn't very sensitive, all the values will remain about the same. If you do this for all the other parameters you'll be able to see which have the most impact. It might also give you confidence (or remove some confidence) in the final conclusion. Say you find the model really isn't sensitive at all to the parameters, and the correlation is always relatively high. Then it doesn't matter that you have some finger in the air assumptions, because even with real values you'd reach the same conclusion. If it turns out to be very sensitive, then maybe finding accurate values is important to really trust it.
It occurred to me as I wrote this reply that with a large enough data set, one could probably estimate these parameters from the data. I say that without thinking through the details, just an intuition that it would be possible. However, that might turn into a pretty complicated inference problem and end up being more trouble then it's worth.
Erm, wow. o.O I'm going to make a coffee now.
Whoa. This is a great article.
I'm reading it again for the second time to soak it in.
look at the profiles and what happens when you link (not from but) to your profile from any of your pages be it on-page or off-page - I'm telling you; it's about profiles and associations, the likes and shares are just background noise, they're part of it as well, but minimal, the value and metrics that count are the on the profiles
Looks like alot of people are math challenged here.
Thanks for the breakdown -
As always it's extremely difficult to understand the effect of just one factor. Links and social are always going to be connected via the page/content so when you look at correlation studies between links and social - how do you rule out this connection?
Take the following scenario: Social sharing helps the (quality) content get discovered which leads to more links and as a result improved rankings links...
Are we talking about a direct or indirect influence?
I've got my magic marker ready...
With the transfer of social equity for the SEO for Excel guide from one url to another url did you test canonicalising the URL to transfer the social attributes, just doing the 301 usually only transfers the Tweets, I have noticed if you want to transfer likes as well you can do it with a canonical.
Will - thank you for an awesome post!
I think I may have to read it again to absorb 100% of the information however I love the science/mathematics behind it. Sports and the lucky-socks were an AWESOME analogy!
I seem to remember from my Psychology degree (one of the few things I do) that there are statistical analyses you can employ to assess correlational/regression data in a way that might offer a causal explanation. Something along the lines of a one way ANOVA using ranking as your independent variable (categorised as first page, second page, third page) and Likes as your dependent variable and then using various values (e.g. inbound links) as a mediator to see what regulates the variation.
Again, it's been a while since I've done stats so it might be way off the mark, but I remember these sorts of tests from my dissertation. Obviously you would need a ton of data, so it's just a thought!
Thanks for this post Will - its good to see some common sense floating in :)
I'm a bit confused since there isn't a conclusive result that could demonstrate the actual effect.
However, I guess none of us will ever know for sure, as in plenty other search engine speculations.
Amazing research!
Just to confirm - are all of the SERP results run here on the API or also natural user search?
Are API search results always consistent, regardless of user?
Good article indeed! Social media platforms have become mandatory places to get noticed that play vital role in higher SERP. Likewise social signals carry more importance in influencing the rankings in Google...
Fantastic info! You've compiled a LOT of insightful and useful data. You either have a superb staff or no girlfriend to occupy your time, lol.
I suspect that Google views certain social signals (interaction from "people", not "websites") as being more genuine/less spammy "approvals" than easily obtained generic inbound links. I think too that there is going to be a similar background assessment of your +1's/Likes, etc. to evaluate how enagegd you are in a particular community. To get "social signals" you truly have to engage and interest people (users) whereas old style links simply require working off of a list of linkable sites or subscribing to a service.
Hah. I do have a superb staff (who helped me proof this post and make it a funnier) but I wrote it. I have a wife, a toddler and a newborn son. I did the modelling on the train to work.
You're right that it probably shouldn't have been a priority, but I got interested and carried away :)
There's certainly a direct causation involved here, the question remains how much of a direct factor does social media play in Google rankings... Someone mentioned an example earlier that lends me to believe that the impact can be rather significant, albeit not as long-lasting as traditional link building.
I believe when asked about the subject Matt Cutts stated that they're a "strong" factor. But, of course, it goes deeper than just obtaining a handful of likes or tweets. The authorship in which the citation came from matters as well, prohibiting people from creating random accounts to spam their links and benefit from it. With that said, I feel it's very important to connect with and foster strong relationships with authoritative sources within your niche across various social media platforms. For the direct and indirect SEO benefits that such a strategy can provide.
A very good article . I think google 1+ will lead to wrong way in the future , because people who want their sites get higher rank in SEO will make a lot of fake +1 , That's against for what google think it will be
We've done significant case studies with our clients regarding social impact on domain authority. I had not seen this article yet though.
We definitely saw a huge increase in all of our clients who've focused on increasing their social signals through tweets, likes, and google plus. While I can see why facebook would have a bigger impact on rankings, we always received way more tweet traffic from twitter users then facebook users. This is just innate to the twitter user culture though I suppose.
With recent changes from Google, local search in Google Maps and Google Plus are one. Hugely important for every company to optimize their Business Google Plus Listing.
We just received a client who accidentally remove his local Google Plus listing (Oh No!). He went from an average of 64 calls every two weeks to 35. While this could have been caused by other factors...his optimized Google plus page played a huge factor.
Thanks for taking this study that much deeper!
What a remarkable body of research. Well done!
Although you come to very interesting results, I have to say that I still believe that Google in particular uses social signals as a signal to judge whether a page should be number 1, 2 or 3 in the SERPs. I totally agree with the correlation, though.
Nice Post, just downloaded the excel model.
Henrique Troitinho
I am sure there is a positive correlation between social signals and ranking. However, what is the mechanic involved with this? Is it only from the rankings we come to know about the improvement? Is there a definite route or a science we can understand?
Please throw some light on it.
Great article...had to read it twice & interesting comments following...
Am very sad there is no answer! LOL Will we ever have one? From our side - we do (feel!) we see a link between our social (Facebook) and our ranking for certain keywords - and ranking for them in the 1st place.
We tend to generate a lot of shares, averaging 95 per post over the last 100 posts and we do "seem" to have a spike in our ranking for these links/keywords we post & have shared.
The challenge is that they 'do not seem' to be permanent ranking boosts - ie in the past 3 weeks we have lost nearly 40% of our links from FB. Strangely some of the links that remained are the same.
We primarily use Facebook in the social world - mastering it before next battle...and of course wanting to measure the impact. Reading the article - while really enjoyable, I was regretting the time we spent on our FB project - but reviewing our own stats - feel that the boost we are getting is enough to continue in our efforts & investment...just wish there was a way to maintain these 'social signals' and the boost we see from them.
Of course - there could be MANY other factors - its always a test & measure - and use your own best judgement on what you see.
Good stuff Will.
I know for a fact, that social signals play a very important role in SEO in our days, it is mandatory now. Social Signals don't only help in SEO but they get you traffic directly as well!
We're using data from Google Webmaster Tools (Average ranking) and our own keyword ranking tools, we are going to run a few tests on articles that are ranking for certain keywords between 5 and 15 position in Google. We are going to run an adwords campaign to these articles to drive likes, shares, tweets etc. FYI, most of these articles only have 2 or 3 social shares at most. If the articles get a lot more shares and rankings improve, this should (in my mind) prove that social sharing as a ranking factor is causation and not correlation.
Your thoughts?
Hi webgrowth,
Sounds like an interesting experiment.
I'm not sure your question about whether you'll be able to prove anything can be answered until you see the data from the tests. If you see an emphatic change in rankings in a short timeframe you might feel it is suggestive of cause and effect but if there is say a moderate change in rankings with any kind of inconsistency in the response then you might be left in doubt.
Also, if rankings did improve, you'd need to eliminate the underlying cause being back links generated indirectly by the social sharing e.g. the social sharing resulted in a few back links from blogs etc.
The great thing about experiments though is that you sometimes get surprising results and learn things you couldn't have predicted you'd learn.
If you assume that one of the primary reasons for the launch of Google Plus was to gain control of social signals (an assumption that I do not have much trouble making), it is easy to jump to a conclusion that social signals are likely a ranking factor. My guess is that it is only a matter of time before someone releases a study that shows that social signals are a ranking factor. Of course, once this study has been published, Google will then change their algorithm, likely invalidating the results
thanks will critchlow, it's been great to read your article and got some interesting fact how our social ca impact more positively on our ranking factors which you have shown quite intelligently . . thanks so much. .yes , now i am going to get this checked !
RambSEO, I interpret the main thrust of Will's message as not so much that social can impact more positively on our rankings but that we should keep an open mind.
I am a new member here i read the article and the comments of others i felt now Social Signals are most important for every SEO organization but i am willing to know which social network gives a better value?
Social signals like Facebook likes, tweets these are good but these activities no longer support to our keyword SERP.
From always backlinks play a crucial role in our SERP and I always see changes in my SERP when I get natural links for my website. Social sharing is good to get traffic on our website but it doesn't create any impact on our ranking.
Hi Will,
A great contribution to a fascinating ongoing debate.
I see your 'punchline' headline as uncontestable. It must be automatically possible until someone provides evidence to the contrary. It is clear from your conclusion that you don't believe anybody has provided evidence and I agree with you.
One thing I don't think I've seen anywhere is a 'common sense' explanation of why social signals should directly result in better rankings. Does social sharing necessarily imply higher relevance or higher authority? If so, I could understand the rationale for a causal relationship. Given two identical websites, businesses etc where the only difference is that one is a big socialiser and the other is not, should the social business necessarily rank better? On a purely intuitive basis, I'm not sure why it should. Are you?
[PS. I understand the socks analogy but I don't think it's a good one. The reason athletes have these rituals and superstitions is to support a belief system required to give them the tiniest edge by helping them to believe they can and will win. These rituals may appear irrational but they are widespread and there is a reason for that. They help. If the athlete has developed an unshakeable belief that, by wearing those socks, he will run faster, then he often will run faster when he wears those socks. It's like the placebo effect and about sports psychology, not statistics.]
I think the argument essentially boils down to the same one used to justify the original PageRank experiments - that people generally share good things.
Many more people have access to social media than have the ability to create a link on a website so it could be seen as more democratic.
[I'm a big placebo fan so I totally get it from that perspective - just that we should realise that it's far more likely that our pattern-matching ability for spotting order among randomness outstrips the likelihood of these rituals making an actual physical difference.]
Thanks for replying Will.
I'm with you on the social.
By saying "... we should realise that it's far more likely .." you're expressing your opinion as though it's fact. Anything to support that?
https://pss.sagepub.com/content/21/7/1014.abstract
Although correlation cannot be easily proven as causation I think from a personal point of view that social data SHOULD be considered as a ranking factor.
For example if I find an article or web page useful I am more inclined to post it via social media than to post it on a forum or a blog where it would get a standard dofollow link. I'm not sure if this is necessarily the case for other people, but I wouldn't be surprised if it is. The engines want to deliver the most relevant results, and a lot of data about what people find useful/relevant is floating around in social spaces.
I liked that little 'Sheldon moment'.
Brilliantly executed experiment. 1) Regarding the loss of social signals caused by the redirection of Mike's article, it is possible and quite simple to manipulate opengraph data and attribute the original count to the new page. You may have done this, but thought I'd mention it. 2) I could be convinced by uploading 2 new websites with similar content on separate IPs, get them indexed and acquire equal inbound links from the same sites for both to kick start some movement in the SERPs. Track movement on both sites while generating social signals for one but not the other.
I like how you pointed out where the social data lies within the spearman coefficient formula falls into place. Cheers!
I've been struggling to get around what I think about this topic until now. Now I know I can just stick to the sentence "there is correlation but no guaranteed causation"
For a given page, with a decent piece of content, people will share it if it's of high quality, likewise google will try to serve it up more often (read: up it's rankings) because people are searching for it more and it is receiving a higher CTR than the others etc etc etc. The "sunny day" causing the "ice-cream sales" and "sun-lotion sales", is that the content is good!
Points to the idea of social signals being part of a "freshness" factor, I like that. hahaha
This backs up my hypothesis, that the higher social metrics are caused by the ranking, and not the other way around. Or rather that more traffic = more likes, and all things equal except search term ranking, the highest ranked page will get more traffic.
With all respect Adam, it doesn't.
It doesn't back up your hypothesis any more than it does the hypothesis that likes are a ranking factor. There is correlation, and reasonable grounds to consider both hypotheses plausible - but your idea is nevertheless still conjecture.
cool article, i do believe that social media has become a strong point of ranking aswell as customer loyalty along with lead generation.
existing on social media is sign of authenticity of any business through which you could stay connected with your customers.
Ahh Dude Lyoto Machidais Winning! lol.. yukk!
IMO NO social, NO results!. Back in the day you could build links via comments/spam/junk/directories/article spinning/booboo... Not anymore those links don't bring value, they are discounted or beter yet penalized. Now'a days NO social = NO links! (likes) and no links = no rankings! Sure we can all create tons of unique/great articles on our blogs but if no one is interacting on the site it will NOT rank.
This post is great thx Will.
As i just said on twitter social means its popular now, it doesn't mean ranking - that is still mainly controlled by quality of content, trust of site and links (plus lots of other stuff) - social is a tiny part of it. don't get me wrong there maybe correlation but it doesn't mean causation as will says
Great and detailed post, not sure I understood much of it, I shall have to read again and again.
Could someone please tell me why Google+ isn't included in studies like these?
I checked the Searchmetrics article at the top of this one as well, and they didn't include Google+ either.
I would think that out of all social signals, Google+ would be the strongest since it's from Google itself.
I mentioned that my analysis works exactly the same with +1s or Tweets. I used FB for my example because that's the signal that people have found the highest confidence in high correlation.
I think the last moz study was done long enough ago that +1s were either too sparse or not even available.
Searchmetrics addressed the question in their write-up though - they said that the data was so sparse that while they found a high correlation, they didn't have high confidence levels.
Thanks for the post - supports some of my personal theories on the topic. Correlation over causality is the safer assumption for now.
It might be as simple as content that gets more Likes is exposed to more eyeballs (through the ticker AND the main feed), thus bringing in more unique visitors and possibly getting shared on more sites/blogs with backlinks. On that note, I would love to see a study on how Likes and retweets affect incoming backlinks.
Looks like you guys at SEOmoz don't have enough clients to deal with, that's why you can invest so much time in such an extensive research.... Just kidding :-)
I always believed that social signals do play a part in the search rankings. With the increased integration of G+ in Google and on the other handBing deepens social ties with Foursquare, it is only going to have more and more effect in the search results.
Thank you for the excellent, detailed analysis. Like many others, I need to read this a few times to let it sink in.
Others have pointed this out, but let me say it again, your example of the correlation between ice cream and sunblock sales with the causation of hot days is a lovely, simple and useful way to frame the concept. :)
What are your thoughts on Google + in the future? If you +1 something that article/page is ranked higher for your friends (I know this is not the true ranking just bear with me for a second) But if one person from every circle +1 a certain article Wouldn't Google see that that content is relevant and important leading to higher search ranking?
Sorry I may be way off with things but I am trying to wrap my head around Google + and since it is their own social media function how it affects rankings in Google.
After penguin update social signal becomes more important for site ranking and increasing visitors, because Google gives more preference to those site which has strong social signals. My site ranking is not dropped after penguin update because i have strong social shares.
Thanks for putting a long running theory to bed Will.
I think your article went over the heads of a lot of people though sadly.
I think the SEO industry needs to do some time series analysis in order to gain any insight in the problem. It simply is not enough to have data at one point of time: we have to have time series of tweets, likes, and rankings.
I've thought this for a while - however I also think at some point or another there'll be a moment when Google just turns it on as a ranking factor and all those that haven't been doing it will drop off, as you say by and large I'm confident there will be a large amount of correlation between the the social media profile of a business and rankings.
Impressive. The concerning about social signals is too big.
As more of a creative SEO with a marketing based background rather than mathematical and technical type, I started to read this article and got scared off by the algorithms. 0_0 As some others have said I really do think it's a simple as Make Good Content, (or sell good products) People will share it
It's interesting to see the increase in likes with the ranking of the site though. I will cut out and keep this in my SEO by Pictures book which as a creative type makes a much easier read
I remember we discussed this topic last year at SMX Paris.
I agree it's very difficult to isolate one factor, indeed.
Moreover, this social thing is better to take as part of a whole strategy, in order to create some kind of emulsion. Social signals alone won't go very far.
Social signals like Facebook likes, tweets these are good but these activities no longer support to our keyword SERP.
From always backlinks play a crucial role in our SERP and I always see changes in my SERP when I get natural links for my website. Social sharing is good to get traffic on our website but it doesn't create any impact on our ranking.
Social likes, mentions and shares can give your content initial temporary boost in rankings but it loose its effect after sometime. Its your content quality and backlinks that will actually decide your rankings in SERP.
Yes, I do believe Social signals are the one of the Important factors in ranking. Thanks for your awesome research
Did you even read the article? Seriously?
Either he hasn't, it he's joking.
I don't know whether Tapan has read the article or not. And Congratulation!!! you are not alone here, 15 thumbs up, means there are many others having same thinking..... :-(
[Jen - edited to remove out-of-context discussion]
Possible troll in your vicinity.
This is why we don't have nice things.
Yes, What Do you think?
I think people should read article before commenting )