It’s Not “He Said, She Said” Over Google Rankings & Facebook Shares
On Tuesday morning, SEOmoz’s Rand Fishkin presented evidence at our SMX Advanced conference that there’s a high correlation between Facebook Shares and ranking well on search engines. In afternoon, the head of Google’s web spam team Matt Cutts said that Google doesn’t see Facebook Share data at all. An SEO-search engine catfight? No, both actually agree.
I’ve been somewhat amazed to read some of the tweets this evening that are positioning what Matt said as somehow contradicting what Rand said. A few on Twitter are even squaring off over whom to believe. That’s dumb, when both “sides” agree.
Facebook Shares & Search Engine Rankings
I moderated our morning panel, The New Periodic Table Of SEO, where Rand talked about findings from the latest SEOmoz ranking factors survey. Facebook Shares, he said, were the top factor that correlated with high search engine rankings. From his presentation:
But Rand never said this meant that Facebook Shares were causing top rankings.
Correlation Versus Causation
Correlation isn’t causation. It’s a mantra that the folks at SEOmoz have learned to stress over the years. Rand’s presentation highlighted this from the start:
I fear that saying this still doesn’t help, because no one really seems to take on the difference. The terms still get used interchangeably.
Both just feel like “dead words” without much meaning behind them. I’m too tired at the moment to brainstorm better words. But let me try to explain the difference perhaps in another way.
If A Bird Flies Past Your Window…
Let’s say that every time you look out a window, you see a bird fly by. There’s a strong correlation between you looking out the window and birds flying past.
But is looking out the window causing birds to fly past? Who knows? You’d have to research in various ways whether glancing out that window was a causation factor making birds decide to fly by.
It could simply be that birds are always flying by the window, so often that every time you look up, you see one. The correlation of you seeing birds isn’t actually causing the birds to appear.
When it comes to search engine ranking factors, researching what is having an impact is incredibly hard, because there are so many variables involved — thousands of sub-factors. It’s easy to mistake a correlation as a causation factor.
So Who’s Right? Both!
Back to today’s “dispute.” In the afternoon, I talked with Matt as part of the “You&A With Matt Cutts” session. Matt made a point to clarified that Google didn’t use any Facebook Share data. He was very specific. Google couldn’t see this data at all, due to Facebook blocking it.
That statement has caused some to view what Matt said as contradicting what Rand said. After all, Rand said Facebook Shares were helping pages rank better. If Matt says Google doesn’t see share data, then someone’s wrong!
The problem with this supposed contradiction is that Rand never said that Facebook Shares were boosting rankings. He simply said that SEOmoz found top ranking content had a lot of shares. It was correlated with that, but that didn’t indicate causation.
Rand was very clear about this. He said several times there was no proof that shares were causing ranking boost. He even said at one point that the shares might have nothing at all to do with the rankings.
Facebook Shares = Quality Content?
So why care at all about knowing such a correlation? For one, as I discussed with Matt, it can be a useful metric for publishers to know if they’re struggling to tell if they have quality content.
Google repeatedly says that it wants to reward quality content, but there’s no quality meter you can wave over your pages to tell if they’re up to snuff.
However, if there’s a strong correlation between pages that get shared on Facebook and pages that rank well on Google, then it could be that shares work as a good proxy for figuring out what Google considers quality — even if Google doesn’t use that share data itself.
In other words, if you’re producing content that doesn’t get shared much, it might not be that great. If it’s not that great to the humans who share things on Facebook, then Google might not like it either. After all, Google’s ranking algorithm tries to mimic what humans like.
We Interrupt This Story For A Postscript!
Since I wrote the story above, Rand’s been conducting a test to see if he can verify if Google is indeed gathering Facebook Share information. We also had some conversation on Twitter about what Google might or might not pull.
Originally, I started to write this postscript as an email to Rand picking up from our Twitter discussion, given that the issues are hard to explain 140 characters at a time. But I thought everyone would be interested, so I’m sharing it here. No pun intended.
It’s also pretty late (6AM ET), so if there are typos, sorry. I’ll fix them later.
What Matt Said
I don’t have the video yet of my talk with Matt, where he discussed the issue with Facebook Shares, so I can’t quote him exactly. But from our live blogging, we have this:
Google doesn’t get Facebook shares. We’re blocked by that data. We can see fan pages, but we can’t see Facebook shares.
From the Bruce Clay coverage:
SEOmoz said Facebook shares in an earlier session helps ranking. Matt says Google doesn’t get Facebook shares. If someone blocks them from crawling, they can’t call that content. Google can see Fan pages but that’s it. The most correlated thing (Facebook shares) is not a signal [Google] uses. If you have great content, it may get a lot of links and that’s no surprise.
From Outspoken Media’s coverage:
If someone blocks us from crawling, we can’t crawl that content. Facebook gives data to Bing, Blekko, etc, but they don’t give it to us. They can see Fan pages, but that’s it. They don’t see the sharing data.
From the paraphrases, there’s the impression that Google can’t crawl Facebook to gather shares.
But It Does Crawl!
Google isn’t blocked from crawling all of Facebook, however. Indeed, it crawls many pages. Facebook wants many of those pages crawled.
Potentially, if you share a link on your personal wall, Google might see that share. After all, it does see a limited version of personal walls. This is one way it can determine some of your Facebook friends. The article below explains more about this:
In that article, however, you’ll see a personal wall shown without any of the shares on it. I’m pretty sure that Facebook isn’t showing shares on any personal walls. If that’s the case, then Google isn’t getting shares that way.
It’s a different situation with public fan pages. Google does crawl these for web search, since Facebook doesn’t block them.
Moreover, Facebook considers everything on a fan page to be public, which means Google and other search engines aren’t shown a limited view, as with personal pages. They see the entire fan page, including all shares on it.
You can see this with Search Engine Land’s own fan page, as cached by Google. All our shares (which are links back to our own articles) are on the page. However, the shares seem to be tagged nofollow, so they might be seen but not actually recorded in Google’s index.
But Rand Got Indexed!
Indeed, when I looked, both Rand’s original share and another one showed up, as shown above. So there you go. Proof that Google indexes Facebook Shares!
Yes and no. Look over to the left, and you’ll see that the results are only showing for Google Realtime Search. That’s a completely separate search engine from Google web search (which is what the SEOmoz study looked at, and what Matt was responding to).
When it comes to web search, so far, the shares aren’t showing:
Now, potentially I’ve messed up Rand’s experiment by linking to his share outside of Facebook. But someone else has already done that, at this point, I can tell from the comments.
More important, Google has a deal with Facebook specifically allowing it to pull in Facebook fan page information. It gets this information directly from Facebook through a feed, to my understanding, rather than having to crawl it. The story below explains more:
That feed probably drops the nofollow attribute, by the way — just as Twitter’s does, which is why you see the shares showing despite nofollow on the public page. The story below explains more about this:
Does The Fan Page Data Flow To Web Search?
Could the data from Google Realtime Search be used to support web search. IE, if Google Realtime Search can see Facebook Shares on fan pages, can those shares then be counted by web search.
Potentially. In fact, Google’s given the impression in the past that this is the case. In our What Social Signals Do Google & Bing Really Count? article from last December, I asked Google:
Do you track links shared within Facebook, either through personal walls or fan pages?
We treat links shared on Facebook fan pages the same as we treat tweeted links. We have no personal wall data from Facebook.
This means that yes, Google sees Facebook fan page data for web search and yes, if those fan pages include shares, it would see that data as well.
All I can tell you is that my original article had been checked and double-checked by Google before it went out. To my understanding, it remains valid. Tweets can have an impact on web search though a much more dramatic impact for Google Realtime Search, Google Social Search and Google Personalized Search.
The articles below cover this more:
- Google’s Search Results Get More Social; Twitter As The New Facebook “Like”
- Meet +1: Google’s Answer To The Facebook Like Button
I think this also means that Facebook fan page data, which includes shares, may have an impact on web search, in very limited cases. But officially, Matt’s saying that’s not the case.
Are All Shares Counted?
Is Matt mistaken? I’ll double-check on this again. Possibly, he meant that Google can’t get Facebook Shares in the sense getting them from where most happen (as I’d guess), on personal walls, rather than on public fan pages.
Certainly the stats that SEOmoz is getting from the Facebook Open Graph for its survey are a mixture of personal wall shares plus public fan shares (see the cool tool Shared Count for an example of this). But Google, at best, is only seeing some of that.
Why can’t Google tap into the entire Facebook Open Graph and see all shares in the way that seemingly anyone else can? Well, Google can — if it wants to agree to Facebook’s terms. And Google doesn’t want to do that. The article below explains this more:
This, by the way, makes Google’s acquisition of PostRank last week very interesting. Potentially, that gives Google insight into Facebook sharing data without Google having to let Facebook see into its actual search network. Of course, if that’s the case, potentially Facebook might cut PostRank off…
Since the PostRank purchase is so new, at best, it seems that Google might be seeing a portion of Facebook shares that in turn could be used in web search. But since it can’t see them all, I’d be surprised if they were using them as a significant ranking signal.
Finally, Rand had pointed out something else that made it seem as if Google was using share data:
What’s happening here isn’t that share data is being leveraged. Instead, Google is using the limited data about friends from personal walls, to my understanding. Again, the article below explains more:
Finally, just a reminder that Facebook Shares are not the same as Facebook Likes. Could Google count Likes?
Likes aren’t links, as with Shares. Likes are associated with shared links or fan pages. Potentially, Google could crawl open fan pages to discover links that are well Liked or pages themselves with many Likes.
Do Factors Really Matter?
And now, back to the last two paragraphs of my original story….
In the end, the debate over whether one can scientifically measure all the factors that a search engine likes or doesn’t like is ages-old in the space. It can be way too easy for those doing surveys, or those reading them, to mistake correlation with cause. It can definitely be way too easy for people to get lost among the various factors.
But looking at the data and combining that with what’s confirmed by the search engines or suspected within the space can also be useful, too. That, by the way, is what our recently released Periodic Table Of SEO Ranking Factors aims to do — focus on the most proven or strongly believed factors. So, be sure to check that out.
- Social Media & Ranking In Search Results
- What Social Signals Do Google & Bing Really Count?
- How Facebook Enables The Google Social “Scraping” It’s Upset About
- Has Facebook Become The Master Key To Unlocking The Web?
- By The Numbers: How Facebook Says Likes & Social Plugins Help Websites
- Bing Ups Ante In Social Search, Adds More Facebook “Likes” To Search Results
- Meet +1: Google’s Answer To The Facebook Like Button
- Introducing: The Periodic Table Of SEO Ranking Factors