Wil Reynolds, founder of the SEO/SEM company SEER Interactive, is on a mission in this special ZDNet guest post: to show that the firestorm surrounding social media's direct impact on search engine rankings is in need of being tempered. Not extinguished! Just tempered. As you will see, Wil has brought a solid argument to the table with some hard data, but he's not suggesting that social media has no impact overall; rather, that in the face of traditional SEO methods, social signals appear to hold very little weight where directly affecting search engine rankings is concerned.
And now, without further ado, here is Wil Reynolds's exposé on SEO rankings.
Wil Reynolds, SEER Interactive
There is no denying, the world is on fire with social media right now. People are talking about how much it is changing the traditional role of SEO, and while it is definitely coming…it's not the be all and end all yet, and I will lay out the data to prove it.
I am not here to extinguish that fire too much; we definitely see tons of ways to use social to drive SEO rankings, but we need to contain that bit and here's why:
All these people betting on social to impact SEO have it right: social is totally impacting SEO, but how much? Probably NOT enough to move the needle on most of the keywords you care about. I believe there are tons of indirect ways social impacts rankings, but that is for another day. Here is why I am not buying that social is impacting SEO in a direct and meaningful way.
I know a ton of amazing SEO companies. They are speaking at conferences, writing great blog posts, connecting with people on social media -- you know, all those things that create REAL relationships with REAL people. Yet god forbid you Google something related to finding SEO companies. 90% of the time, the SEO companies out there putting in the hard work of writing blog posts, connecting with people on social, and leading our industry never show up and I have the data to illustrate this point.
Let's get started with a very telling graph:
What will shock you about the data above is that companies 5-7 are some of the hands-down leaders in our industry while sites 1-4 are far from it, yet by getting more links with the word SEO in them, they are outranking the quality sites, which you’ll see below. Isn’t it shocking that a company with 80% of their links having a keyword in it is not a trigger to Google that they may have an unnatural profile? It's my belief that quality companies are much more likely to have a mix of links with their brand names, some images, and some anchor text.
Let's get into my methodology. I picked blog metrics as an indicator of where Google fails to use social as a strong signal due to the very real fact that outside of Twitter and Facebook, they should be able to see shares and consumption of content better than anyone. Here are some of their own properties that for years had social share and consumption data:
• Feedburner was launched in 2004 and acquired by Google in 2007, meaning, at worst, Google has 4 years of RSS feed consumption and sharing data sitting in their data warehouse. Best case, 7 years.
• Google Reader was launched in October 2005, meaning they have 6 years of data on blogs, who is reading them, sharing info, etc. See my Google reader history below; they HAVE this data, and you can see it for yourself in your Google Reader.
• Google now owns Postrank (launched in January 2007), which only adds to the data they will have sitting in their warehouse to determine the quality of blog posts and engagement/sharing.
• Obviously, Google has a crawler that can see which sites are getting links, when those links are attained, who is linking to a site, what that content says, etc.
This is a pretty powerful combination of assessing the quality of blogs and those who are reading those blogs. Feedburner has access to the following quality data:
• Subscriber growth
• Subscriber retention
Google Reader has access to the following data for readership, letting them know who is real and who is not (read this piece on how creating fake accounts could sabotage social signals):
They also see shares, starred posts, etc. Google has all of this data! Yet companies investing in “connecting” with people by producing quality content -- but who don’t build significant anchor text links -- are likely to not rank well in the current environment.
Postrank (a recent Google acquisition) provided a score based on how “socially promoted” one’s blog posts were on Facebook, Twitter, Delicious, and other social sites. In spite of having access to this level of “quality” blog posts in their data warehouse, Google obviously hasn’t been able to include it with a high enough weight in the algorithm to have an impact on overall rankings. Just goes to show how complicated managing an algorithm is.
I figure a picture tells 1000 words, so I decided to create some telling graphs to give ammo to my argument. Sites 1-4 are what I would call "high ranking sites" for SEO company queries with no real recognizable name in the industry. Sites 5-7 are recognized names who were not ranking on page 1 for several SEO company-style keywords. #7 is SEER Interactive, my SEM/SEO company.
Note: I purposely distorted the numbers a bit and kept the companies anonymous.
Let's start with the easiest metric: Links. If you are constantly writing blog posts (and all sites had blogs of varying degrees) – if those posts are quality, logic tells us that at least some of them should be attracting links, right?
You’ll notice a pattern as you read below, but as you can see above, the sites with the most social activity, inbound links to content, etc. are rarely the ones ranking well.
Let's see how things look when examining how many blog posts have at least 10 RTs (retweets). You can see the sites on the right are well above the others (I know this is a potentially spammable metric, but it is just one sample of social sharing).
When examining the number of blog posts that attracted links with a Domain Authority (what is DA?) over 40 and 55, you can see the companies on the right side of the graph again outpace the others by far, yet they are NOT rewarded with rankings for keywords like "SEO company" or "SEO firm."
I picked this metric because Google has a value for every site on the web and it is constantly calculated and re-calculated over and over again. We obviously don’t have access to that, so we are using the SEOmoz MozTrust figures in the absence of real Google data.
Links to your site from other sites with high trust signals (NYtimes, PC World, Industry Publications) are very difficult to get. The higher the % of links with high Page/Domain Authority, the more “trusted” your links are.
Branded anchor text
It's obvious that we all need anchor text links to help us rank well. It is equally as obvious that a company that doesn’t get any links to its web site using the company name is probably none too natural. Google can obviously see this with its existing technology. Getting anchor text with your keywords in it is RARELY the natural way to link, yet if you don’t do it, you don’t get rewarded with rankings. I look forward to the day when this is the case.
In the example above, you can see that the “higher quality” companies have a higher percentage of links with their brand names as the anchor text (note: none of the sites include the word SEO as part of their domain name).
Then there are the non-Google figures they can partner with to see if a site ranking well is sending social signals -- one example being Twitter. Below you will see followers per day (pulled by Followerwonk).
As you’ve seen time and time again, the "high quality" companies are bunched to the right, yet they are not ranking well. Followers per day is an easily spammed metric as well, which is probably why it shouldn’t influence rankings. But with the whiz kids at Google, I imagine they could add some quality metrics to followers as a way of seeing if you are attracting fake profiles or real true users.
I put all of this together to say... Google has access to all this data in their own data warehouse and could easily partner with other companies to create a more holistic picture of web properties doing real social engagement. And while they may have created that holistic view, as you can see, it is not heavily impacting the algorithm for any remotely competitive keywords. I speculate that might be what the Postrank acquisition was all about.
I personally cannot wait for the day when SEER Interactive and a LOT of companies across every industry who have invested heavily in doing things right get recognized. Many SEO companies like us are speaking at conferences, writing guest posts, getting listed on high authority sites, attaining news mentions in quality sources, connecting with customers on sites like Facebook and Twitter, and writing great content that people actually want to read and share -- yet right now, doing all of that is such a small part of how to succeed on Google, that there is no impact. We are waiting, Google. We can’t wait for you to get this right.
For more from Wil Reynolds, be sure to check out his company, SEER Interactive. Additionally, you can find him via the following social avenues: Twitter, Delicious, YouTube, LinkedIn
-Stephen Chapman SEO Whistleblower Related Content:
Stephen is a freelance writer and blogger based in Charlotte, NC. His contributions to ZDNet cover topics related to security, gaming, Microsoft, Apple, and other topics of interest with a tech/SMB skew.