SEOHow Much Do Social Signals Play Into Google Rankings?

How Much Do Social Signals Play Into Google Rankings?

When it comes to the big black box that is Google, we will never know exactly what's going on. Does Google use social signals as a ranking factor? A lot of folks seem to think so. Are they wrong? Or is the answer a bit more complex?

If there’s been one thing that tends to get me going, it’s when people in the industry start talking about social signals. In particular, Google using them as a ranking factor. Is this truly the case? Or do a lot of folks have it wrong? Or is the answer a bit more complex?

That’s what we’ll get into today.

For starters, let’s get the easy bits out of the way. Most social websites of consequence use the nofollow attribute. This means that the traditional PageRank approach to scoring is a non-starter. Yes, there are those that believe Google is selective in how they treat NF links, I am certainly not one of them.


Personalization & Social

One area to consider of course is personalization. We all know by now that when you’re logged into Google these days social elements most certainly can and do, change the rankings and what’s shown for a query.


But that is happening because of the relations one has in the social realm. The question at hand today is more about non-personalized search rankings. The concept that a lot of sharing (legit or not) can affect how the core rankings are scored.

Let’s consider this study Google did about sharing on Google+. They state:

67.6 percent of all items were shared using Circles and 33.8 percent of items were shared publicly (these percentages add up to more than 100 percent since some users combined sharing options for posts).


Here’s the qualitative end of the study, from asking users:

Self Reported Use of Google Sharing Options

They do indeed have a deeper understanding of sharing that in the past when only accessing publicly available data. It stands to reason that they may see sharing as a more personalized activity than a broadcast channel. In that study they also look at naming conventions for Circles where they consider categorizations of ‘life facets’ (associations such as work, school etc..) and ‘tie strength’ (is a circle a catch-all or inner circle?).

You can read the study yourself, but my point is not so subtle: social signals are possibly being seen as personal.

Signal vs. Ranking Factor

We next need to consider that a signal isn’t always the same as a ranking factor. Google has said, (in late 2010), in regards to retweets:

“Yes, we do use it as a signal. It is used as a signal in our organic and news rankings. We also use it to enhance our news universal by marking how many people shared an article.”

By and large social elements are for display, more so than as a scoring element that re-ranks results. An important distinction.

Other types of signals can include:

  • Discovery
  • Trust concepts (pages and users)
  • Temporal (velocity)
  • Context (semantics)
  • Behavioral

Of the above, the more sensible are discovery and velocity. Certainly it can be said that a page can’t rank if it isn’t indexed, so social channels can have an effect as far as discovery is concerned.

We also have elements such as the QDF (query deserves freshness). There could be some element of social sharing velocity that could play into that. I’ll grudgingly give that some consideration.

As for potential social signals that could be used for search rankings, Bill Slawski highlighted a post about Google’s acquisition of some Groupivity/Appmail patents. Some types of implicit/explicit data he highlighted include;

  • Sending (via email, instant messenger, etc.) a link or bookmark to content to another user
  • Posting content (e.g., adding to a blog which becomes visible to others)
  • Applying a label to content visible to others
  • Making making a purchase or request from a website
  • Participating in social bookmarking
  • Promoting or demoting content in search results
  • Posting comments on bookmarks, news, images, videos, podcasts and other web pages

All are sensible approaches, but they do seem to make more sense in a personalized (logged-in) state than for open search. These forms of behavioral data would make tailoring content more robust and may be too noisy in a normal search setting.

Authority & Trust

Now, if I were to start to consider social as a ranking factor, I’d be inclined to look at how search engines deal with entities and trust. Meaning, the entity that is sharing the content would need to be a trusted source for any search engine to truly use it as a type of (non-personalized) scoring factor.

This is where it starts to make more sense. Social media signals are inherently noisy and spammable. By giving different weights based on known entity data, the lower end of the spectrum would have less, or no effect on search rankings. That can help with spam to a large degree.

Slawski also mentioned the recent addition to AgentRank (patents) doesn’t treat all endorsements, (social votes) the same. Saying that, “those endorsements don’t impact the content being endorsed directly, but rather indirectly by impacting the reputation score of people being endorsed. Which certainly makes sense. He added that, (…) if you endorse something that isn’t worth being endorsed, your reputation score might take a hit.”

It becomes about the quality of your interactions, the value of contributions and sharing. In a situation such as this building one’s reputation in a social graph is considerable effort, whereas losing it can be easy. Just always try to consider your TrustRank concepts; good links to good, and crap often links to other crap.

Another element is of course categorization of users topically and establishing authoritative topics for each. Looking at some of the so-called ‘FriendRank’ patents from Google, they talk about categorizing elements of users within the social graph into;

  • Open Profile (user identification) – Uses identification and scoring/categorization analysis of a user profiles
  • Custodian profile (content relational) – Looks at inferences between the viewer of a web page and the person that created, or manages it.
  • Relationships and Topics – For looking at common relationships among users and related topical categorizations (and behavioral metrics).

Much of this seems to certainly show an interest in more refined social signals and social graph targeting. But still we’re left to wonder; is Google using them beyond personalization?


Correlation and causation

Recently my friends over at Searchmetrics put out an interesting study that looked at some ranking factors and came to the conclusion (as have others) that social activity correlates well with Google rankings.

The study was fairly broad in scope, something I usually bitch about, (thin data sets in the space). They covered some 10,000 keywords, 30,000 SERPs, 300,000 titles, 338 million Facebook comments, 8.1 billion Facebook likes, well… you get the idea. Massive.

Here are some of the findings:


Of course, we have to state the obvious: correlation is not causation necessarily. Slawski said he’s been “a little wary of correlation studies because they sometimes don’t tell us things that aren’t fueled by confirmation bias or coincidence.” In short, unless we actually knew all the factors involved in ranking on Google, it becomes hard to establish exactly what’s going on.

That being said, we can surely say that regardless of Google, engaging in social channels would certainly have a direct or indirect effect on the world of search optimization. In many cases the activity results in wider visibility which can lead to links and other signals (as discussed earlier).

What’s Google Saying?

Google has said in the past that they “(…) do compute and use author quality and that Author authority is independent of PageRank, but it is currently only used in limited situations in ordinary web search.”

Traditionally social signals were used by the (now defunct) Realtime Search and Blog teams at Google. But the last we’d really heard on the matter (from Mr. Cutts) was that they were “studying how much sense it makes to use it a little more widely” in web search rankings.

More recently though on their own sharing system, Cutts has stated that +1s aren’t the best quality signal right now.

If I had to take a guess as to what it all means? Again we can see that Google seems unconvinced on where these signals value lies. If they’re unsure of the value for their own metrics (+1), one has to wonder how comfortable they are with 3rd part signals such as Facebook, Twitter or other social services.

The Final Word

And as always I am sure you’re sitting there (having scanned to the pay-off) wanting to know the question we started with: How much do social signals play into Google rankings?

Don’t be daft, I haven’t a clue. The destination of this excursion was one that leaves you to ponder this yourself. When it comes to the big black box that is Google, we will never know exactly what’s going on. Don’t believe all that you hear. Don’t jump on each passing bandwagon that meanders down the lane. Think about it first.

To rank based on tweets, Likes and +1s may not make a whole lot of sense. Does one count a share or a like? What if those change? Anyone remember that before +1s there was Buzz? How about the Twitter firehose getting turned off? Oh look, a new site called Pinterest! See where I’m headed?

It makes more sense to look at the users (entities) in the social graph instead of the actual sites where they’re active. Search needs to be scalable and one would have to imagine that reliance on any third party site, metric or specific instance (tweet, +1, like etc) seems a bad choice.

Should you stop all your SEO endeavors and just spam social and links? Probably not a good idea. Ultimately there are many reasons to embrace social media in your marketing endeavors. Most of them have little to do with SEO. Embrace it because it makes sense to your business, not because it has magical ranking powers, and you’ll make out fine.

Until next time… play safe.


The 2023 B2B Superpowers Index
whitepaper | Analytics

The 2023 B2B Superpowers Index

Data Analytics in Marketing
whitepaper | Analytics

Data Analytics in Marketing

The Third-Party Data Deprecation Playbook
whitepaper | Digital Marketing

The Third-Party Data Deprecation Playbook

Utilizing Email To Stop Fraud-eCommerce Client Fraud Case Study
whitepaper | Digital Marketing

Utilizing Email To Stop Fraud-eCommerce Client Fraud Case Study