How far along is Google from using social media identities as ranking factors? Can Google use engagement and follower metrics from Twitter and Facebook to evaluate the authority of an individual?
To me, the answers to those questions were the buried headlines in a Google Webmaster Help video by Matt Cutts. Scroll below the video embed to get my commentary and thoughts.
Are Facebook and Twitter Signals Part of the Google Ranking Algorithms?
That’s the question that Matt Cutts chose to answer in this video. Let’s break down the main ideas in his answer.
1. Facebook and Twitter Pages are treated like any other web pages for search. Tweet this!
First, we should understand here that when Matt says “pages” he’s referring to individual pieces of content on those social sites. So on Twitter, that would be a tweet. To Google, each individual tweet is a web page on its own. On Facebook a “page” would be any status update, reshare, link share, etc. you might see in your news feed. Each of the individual “cards” you now see in your news feed,whether from a friend, a Facebook Page, or a group, are each a “page” to Google.
2. Google can only index Facebook and Twitter Pages it can crawl. Tweet this!
Matt made it clear that Google isn’t always able to crawl all of the pages on Facebook and Twitter. In fact, he shared that they had one experience where they were blocked entirely from crawling one of those sites (Barry Schwartz says it was Twitter) for about a month and a half.
The fact that they could get blocked makes Google’s algorithm engineers jittery. They have to worry that they could get blocked again in the future.
Something Matt didn’t mention here, but has elsewhere: even when Google is not blocked from Facebook or Twitter, it still sees an incomplete picture. Neither of those sites currently give Google access to their “fire hose” feeds that provide all of their content. So at the best of times, Google knows it misses a certain amount (perhaps a lot) from those sites.
Google doesn’t like signal sets with big holes in them.
3. Google does NOT currently use signals like Facebook or Twitter followers for search ranking. Tweet this!
At least, as Matt said, to the best of his knowledge. Why? Because Google won’t use a signal to influence its search rankings unless they have high confidence in the meaning of that signal. If Google can’t see all the connections and internal signals about content on a site, then they can’t have that kind of confidence.
In other words, because Google can’t completely crawl Facebook and Twitter, it inevitably is missing lots of data that it would need to do an accurate evaluation of the relative authority of pages within those sites.
Matt gave some examples of the problems that could occur if Google did try to use signals from those sites to rank content (and presumably, individuals, as we’ll get into below).
Problems can occur because social sites by their very nature are volatile. Numbers and relationships change constantly. Google visits each part of the web at “finite moments” as Matt put it. They only see what is happening on a web page at the moment the Google bot visits it. Then the crawler bot moves on, and may not revisit the page for some time.
Say someone had a certain graph of followers at the moment Google crawled their profile, but then shortly after that they did something that caused them to be unfollowed or blocked by a large number of followers. Or a relationship status could change. When you combine the facts that Google only periodically visits a site with how quickly things can change in social media, along with the aforementioned problem of Google getting throttled or blocked from these sites…well you can see why their signal confidence would be low.
4. “Because we’re sampling an imperfect web, we have to worry a lot about identity, when identity is already hard.”
I wanted to get that point as an exact quote. It’s what I’ve been trying to explain to many people with whom I interact who are convinced that Google must by now be using identity authority based on social signals as a ranking factor (for example, the much talked about “Author Rank”).
Getting a usable, high confidence signal based on social profile identities is much more difficult than most people think. I’ve noticed that whenever Matt Cutts and other Google reps have talked about the topic, they use highly qualified language, such as “we want to work toward” of “we are getting better at” assessing authority based on an individual and then using that as a ranking signal.
Cutts has gone so far as saying, in hypothetical examples, that Google using author authority could be as much as ten years off! I don’t think he’s making a literal prediction when he says that. Rather, I believe he’s saying to us: “This is still a long way off.”
5. Social signal correlations with higher rankings for sites do not equal causation of those rankings. Tweet this!
Matt went on to make clear something that caused a fair bit of uproar online in 2013. Several sites, most prominently SearchMetrics and Moz, published correlation studies that showed social signals such as Facebook Likes as one of the highest correlating factors for sites that rank highly in Google search. This caused many bloggers and social media commentators to jump to the conclusion that these social signals were a cause of the higher rankings.
Cooler heads (such as Moz’s own Cyrus Shepard) then tried to explain that a correlating factor doesn’t have to be a causal factor. The more likely explanation, given by Matt Cuts at SMX Advanced in 2013 and repeated here in this video, is that sites that tend to get high social engagement also tend to be sites that are so excellent that they also attract many other signals (such as links) that do actually contribute to search ranking power.
6. Be on social media not for search rankings but to build up your brand and drive qualified traffic. Tweet this!
According to Matt Cutts, there are very valid reasons for being active on all forms of social media even if social media, for now, doesn’t have much or any effect on search rankings.
An active social presence combined with good network building can be a major contributor to growing a brand reputation, better customer service, developing trust and authority, as well as bringing traffic to your sites via the links you post. Those considerations should all be part of any good digital marketer’s arsenal.
7. Understanding identity and social connections for ranking purposes is a long term project. Tweet this!
This section (starting at about 3:20 into the video) is so important!
Some commentators have been quick to jump on any hint from Google that individual authority as a ranking factor remains something Google is “working on” as “evidence” that “Author Rank must now be active!” On the other hand, those of us who have carefully followed this topic of author ranking from the beginning, such as AJ Kohn and myself, have cautioned all along that this area of identifying and rating authors by the perceived trust and reputation of their content is way more complicated than most people suppose.
While no one would like to see subject-area authority for authors become a rankings booster more than I (heck, it would probably help me!), I am not going to declare it “on” just because I wish it were.
Conclusion: Assessing Social Authority Is Like a Fine Wine
As I said above, Google is very careful with their search results. There is no incentive for them to rush an incomplete and unreliable signal into their ranking factors, and plenty of disincentives. Accurately measuring and evaluating the complex signals that might indicate how authoritative an individual or organization is on social media, and especially in regard to specific topic areas, is hugely complex, and made all the more difficult when major areas where such signals exist are difficult for Google to access.
But that does not mean that Google does not value such signals. Every indication we’ve had from Google spokespersons, including Matt Cutts in this video, has been that the areas of social signals and author’s as subject authorities remain areas of intense interest for Google’s engineers.
In his Pubcon Las Vegas 2013 keynote speech, Matt said that social signals should not be looked at for “short term” benefits (i.e., as a direct ranking signal) but rather as a “long term” play. In other words, over time Google will be watching to see who consistently gets good social signals day-in and day-out as an indication of who should be trusted.
Years ago, actor and director Orson Welles was featured in a series of wine commercials on TV that became pop culture memes long before the web. I’m referencing those ads in my own meme image just above. In the ads. Welles would pontificate on the extraordinary efforts the winemaker he was hawking went to to take the time to ensure the quality of the product. He would then turn to the camera and let us know that this wine maker “will serve no wine before it’s time.”
In like manner, Google will serve no ranking factor before its time. Social signals are important, and active use of social media for marketing is now essential, but invest in them for the long term, knowing that if you build real value that people value, over time that will become valuable to Google as well.