Joy Hawkins asked another excellent local SEO question in the Local Search Forums the other day. She asked why do some business listings get a local one box result, while others return the three-pack?
The example she gave was searching for [State Farm Boulder CO] gives you a local one box. Here is a screen shot:
Compared to searching for [state farm boulder] which just gives you the three-pack. Here is a screen shot:
She asked, does having a ton of good reviews give you the one box? The example above, that company has tons of awesome reviews. But most of the local SEOs there say no, reviews are not what give you that one box.
Some believe the one box is based on other SEO factors such as your links and PageRank. While other SEOs say it has to do with how specific the query is. The more specific the query is and the more it matched your business, the more likely you will see a local one box.
I think the one boxes for generic terms are way more unlikely to show than the three pack. Searching specifically on the company name is when most of the one box results show. But to come up for a generic search term in a one box, I guess that takes some fancy SEO work.
Post updated September 1, 2015 and June 7, 2016 (originally published January 23, 2014)
How far along is Google from using social media signals 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 (embedded below) by Matt Cutts. Even though Matt is currently on an extended hiatus from his job as head of Google’s web spam team, I believe what he had to say in this video remains the case today.
My purpose in this post is to examine Matt Cutts’ comments in great detail in order to understand why Google does not incorporate social signals as a ranking factor.
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 posts are treated like any other web pages for search.
First, we should understand 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.
Not all that is indexable is indexed
But what’s really important to understand here is the unspoken implication. Most people assume that Google tries to index every page on the web. Not true! Although Google’s resources are incredibly vast, they do have their limits. Furthermore, with the number of web pages increasing at exponential rates, Google realizes not every page on the web is equally valuable, or even valuable at all to anyone. So they build into their crawling bots algorithms that help them to be selective in what to crawl and how much.
The implications of that are even more profound when it comes to social media, which now churns out many more pieces of content per day than traditional web pages ever could. The number of tweets per day is now well over 500 million!
So it is a safe bet that because of sheer volume alone, Google doesn’t attempt to index all (or even most) of the social posts generated.
2. Google is limited in how much of the social web it can crawl.
Furthermore, 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.
Of course, since Matt posted his video Google and Twitter brokered a new deal that gives Google access to the Twitter “firehose.” That means Google can see every tweet posted in real time. However, for the reasons given in point one above, that doesn’t meant they do index every tweet. In fact, in a study we conducted we found that as of June 2015 Google was still indexing less than 4% of all tweets.
The bottom line: 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.
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 doesn’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 Googlebot 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.”
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.”
Why? Because it’s very difficult to ascertain individual identities across various social platforms. How do I know that the John Smith I follow on Twitter is the same John Smith I see on Instagram?
5. Social signal correlations with higher rankings for sites do not equal causation of those rankings.
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 and Google +1s as one of the highest correlating factors for sites that rank highly in Google search. This caused many to jump to the conclusion that these social signals were acause 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.
Also, increased social media exposure increases the opportunities that sites will link to your content.
6. Be on social media not for search rankings but to build up your brand and drive qualified traffic.
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.
This section (starting at about 3:20 into the video) is so important!
During the three year experiment that was Google Authorship, one of the hottest topics in the SEO world was “author rank,” the idea that Google might use (or be already using) the individual authority of authors for given topics as a search ranking factor.
But in this video Matt Cutts makes clear why utilizing authority of individuals as a ranking signal is a goal for Google, but it remains a long term goal. As mentioned above, establishing and verifying identity of individuals on the web is hard. Add in the reasons given above for the difficulty of assessing social signals, and you can understand why this is not something Google can just turn on like a switch.
