In today’s post we are going to review the data we (Stone Temple) have assembled on the potential for Facebook actity to impact Google rankings.This is as a follow up to the prior study we did to evaluate if Google uses Google+ Shares as a ranking factor. Let’s dig in!
Google Visibility into Facebook Likes
So while Google can load the total number of Likes a page has, it cannot evaluate the quality of those Likes, making the information useless. In fact, every single one of those Likes could have been purchased on Fiverr.
Back in September we published a study where we attemped to directly measure the impact of Google Plus Shares on Ranking. This study showed that there was no clear evidence that the shares in isolation impacted non-personalized search results.
At the time we also ran two tests related to Facebook. One drove a large number of Likes to brand new pages on established domains, and the other shared brand new pages on established domains via Facebook. In both cases, the Liked or shared pages had no links to them, including internal links, and there was no other means for Google to discover the pages other than Facebook.
In the Like test, we obtained a large number of Likes to 2 different web pages on 3 different domains, for 6 total test cases:
As you can see, we drove a considerable number of Likes to each page. About 50 or so from real people we knew, and the rest from Fiverr. The result? None of the pages were ever crawled or indexed by Google. None.
Does Google Index Facebook Shared Content?
Great question. You can intuitively sense that this may have more value. A share implies more commitment by the person doing the sharing, and public Facebook pages and profiles can be crawled by Google. We attempted to test this as well. We asked over 50 people we knew to share a different set of test pages, but as you will see in the screen shot below, we had limited success:
As with the Like test, there was no sign of the pages getting crawled or indexed. However, the number of signals (shares) was small, and the profiles, while real, and public, were not highly authoritative. This may have led to the pages not getting crawled or indexed. This is what led to our performing the next test, which is described below! But first, an analysis into Google’s visibility into your Friends.
Google’s Visibility Into Your Friends
Of course, this is only relevant to public profiles, but if your profile is public, Google can see your Friends link. Here is how it looks in the source code for your profile page:
If you click on that link and go to the next page, on the initial load of the source code via a laptop/desk top computer, the page will show 20 of your friends, and that is all. Here is a sampling of people from my profile:
However, on the mobile version of the Facebook site, there is a clean link on the Friends page to “show more friends”. This is a simple text link and it is not scrolling dependent. We can’t conclusively decide that Google is pulling this data, but clearly if they are motivated to get this data they can. For that reason, for purposes of this article, we will assume that they do!
Thanks to Aimclear’s CTO, Joe Warner, for his help with the analysis into Google’s visibility into a user’s Friends!
How Likely is Google to Index Facebook Posts?
Another great question. The reason it is interesting is because they need to index the post for it to potentially be a ranking factor. Well they certainly index a LOT of Facebook posts:
1.87 billion indexed posts sure seems like a large number. But, we don’t know how many Facebook posts there are, so we decided to measure indexing another way. The approach we used is that we looked at a large number of posts (40) across 85 highly prominent Facebook profiles. These are profiles for people who have large numbers of Likes and a strong PageRank for their page, and are therefore more likely to be seen as authoritative. For each profile we looked at the following:
- Their last 10 posts
- 10 posts that were 3 months or older
- 10 posts that were 6 months or older
- 10 posts that were 12 months or older
In addition, we separately tracked whether or not the posts were simple text posts, posts that included a link, posts with one or more images, or posts with videos to see if that has any impact on Google’s indexing behavior.
The results were fascinating to say the least! Of course, it must be pointed out that 340 posts represents a relatively small number of posts. However, we deliberately picked more prominent profiles so they would be ones that Google was more likely to pay attention too. Here are the results!
Indexing of the Last 10 Posts
The most recent 10 posts are interesting because they should help us understand if Google is treating Facebook posts as an indicator of news. If it is, it should index these quite heavily and quickly for that matter, but that does not seem to be the case. Here is a summary of the data.
Note that the additional 4 pie charts are classifications of the posts, as follows:
- “Images” indicates that a single image or an album was shared in the post.
- “Links” indicates that the post includes a link in it. This is interesting as it could be used as an endorsement in the same fashion as a traditional web link.
- “Text” is for those posts that consists soley of text.
- “Videos” as you would expect is for those posts that contain a video
Indexing of the 10 Posts that are Three Months Old
The following chart, and the next two charts after that, are intended to see if there are material changes in indexing over time, starting with posts that are 3 months old:
Indexing of the 10 Posts that are Six Months Old
Indexing of the 10 Posts that are One Year Old
Indexing of All Posts
Finally, we summarize all the posts we tested:
For those of you that want the gory details, you can see the full spreadsheet here. I am not promising this spreadsheet is easy reading, so be prepared for a time commitment to figure it out, but I offer it up for full disclosure purposes!
Let’s keep this simple and offer a summary of conclusions:
1. Clearly Google does not use the Like data. They don’t have it, except for raw aggregate numbers to which they can attach no value. This conclusion is supported by the test we did to see if 800 or more Likes attracted Google’s attention. It did not.
2. Our Attempt to Directly Measure Indexing and Ranking of Facebook Shared Links Was Inconclusive. We just did not have enough participation to be sure of the results. However, the data we did have showed that Google did not even crawl the pages based on Facebook shares. This is an indicator that Google does not use this either, but we can’t take that one to the bank.
3. If Your Profile is Public, Google Can See Who Your Friends are if They Want. They need to crawl the mobile version of Facebook to do so, but it is certainly available to them.
4. Google Does not Index All Shares on Prominent Profiles. Even for megastars, Google’s indexing of their posts is inconsistent. Under 60% of all their posts are indexed. Interestingly enough, they do index 85% of the posts that contain links, but still not 100%.
5. Query Deserves Freshness Behavior is Not in Evidence. There is no material evidence to show that posts are more likely to be indexed when they are new (and then dropped from the index later). The only interesting exception is images.
My net conclusion – Google doesn’t use Facebook as a discovery, indexing, or ranking factor.
You can, of course disagree with me. Some of you will. Tell me what you think in the comments!