We all know that the one constant in SEO is change, and it can be argued that search engine optimisation is now undergoing one of the most significant periods of change in its history. When Google rolled out Search Plus Your World earlier this year, and it emerged how much Google Plus would impact the SERPs both visually and in terms of actual ranking position (albeit mainly through personalisation). However, it has been suggested that this has come at the detriment of the quality and integrity of their search product through the unnatural promotion of Google Plus profiles in results. However, this seems to have been curbed slightly in a recent set of algorithm changes, one of which was an improvement to how they index public social media profiles. Luckily, Mark Zuckerberg's empty G+ profile no longer ranks first in a search for his name….

Some SEOs have been predicting this for a while now, given how skewed the authority of links has become due to spam and the fact that they are bought, sold and manipulated by those trying to make a quick buck on the internet. The search engines needed a new external metric to gauge the quality of a website, and social engagement seemed like the perfect fit. Like link building, social engagement represents a public endorsement of a website. Like link building, Google can assess the quality of the social sharer, as well as the quantity of social shares a site or piece of content receives. Unlike link building, social engagement isn't so easy to manipulate (not just yet anyway).

Anyway, my names Rory, and I work for SEO Consult, and we thought it would be a good idea to conduct a study into the quantity of socially active traffic that a selection of our client sites receive. Our intention was to reveal some actionable data about the potential for social engagement and the importance of promoting a social presence on-page. Here is what we did:

Social Engagement Experiment Methodology

The results of our experiment are based on the findings from a total of 28 client websites. We selected sites from a wide range of industries, with varying monthly traffic quantities and different target markets (about 60% were B2C targeted, and 40% were B2B targeted). We felt this approach would give us a natural, diverse data range that wasn't unnecessarily broad.

To generate the results we added the following snippet of JavaScript code (inspired by Mike Cardwell at Grepular, and Tom Anthony here at SEOmoz) to the client websites that we would be monitoring for this experiment:

<script>function analyse(b,c){var a=["_trackEvent"];a.push(b,"Signed in?");a.push(c,1,!0);_gaq.push(a);}</script>

<img alt="Flickr" onerror="analyse('Flickr','Out')" onload="analyse('Flickr','In')" src="http://www.flickr.com/signin/yahoo/?redir=%2Fapple-touch-icon.png" style="display:none">

<img alt="Google" onerror="analyse('Google','Out')" onload="analyse('Google','In')" src="https://accounts.google.com/CheckCookie?continue=https%3A%2F%2Fwww.google.com%2Fintl%2Fen%2Fimages%2Flogos%2Faccounts_logo.png" style="display:none">

<img alt="Google+" onerror="analyse('Google+','Out')" onload="analyse('Google+','In')" src="https://plus.google.com/up/?continue=https%3A%2F%2Fwww.google.com%2Fintl%2Fen%2Fimages%2Flogos%2Faccounts_logo.png" style="display:none">

<img alt="Pinterest" onerror="analyse('Pinterest','Out')" onload="analyse('Pinterest','In')" src="https://pinterest.com/login/?next=https%3A%2F%2Fa248.e.akamai.net%2Fpassets.pinterest.com.s3.amazonaws.com%2Fimages%2Flogin_icons.png" style="display:none">

<img alt="Twitter" onerror="analyse('Twitter','Out')" onload="analyse('Twitter','In')" src="https://twitter.com/login?redirect_after_login=%2Fimages%2Fspinner.gif" style="display:none">

<script defer id="Facebook" onerror="analyse('Facebook','Out')" onload="analyse('Facebook','In')" src="https://www.facebook.com/profile.php?id=100002522266847"></script>

The experiment worked like this:

• The URLs for the images’ sources are login ‘doorway’ pages for the respective services that use redirect or referrer URLs in the query string to take the visitor to where they want to go, once they have logged in.
  http://www.doma.in/login-page.html?where_next=http://www.doma.in/an-image.png
• If the visitor is logged in, they will pass though the login ‘doorway’ and be redirected to URL in the query string.
• In this example, they’re redirected to http://www.doma.in/an-image.png.
• If the visitor is logged out, they are served the login ‘doorway’ page.
• If the image element gets served, the image at http://www.doma.in/an-image.png fires the ‘onload’ event, because it has successfully loaded. This triggers Google Analytics, which records the visitor as being signed into a specific social media account.
• If the image element gets served the login ‘doorway’ page the it fires the ‘onerror’ event, because it was expecting an image, and got HTML instead. Again, this triggers Analytics, but records the user as not being logged into the social media account.
• Facebook worked in a similar way because it automatically serves a 404-response code within the header if you visit the hidden profile we created, and are not signed in to FB, and a 200-response code if you are. This is recorded in analytics in the same way as the other social networks.

The Results:

Facebook Stats:

Users signed into Facebook      38.77%
User signed out of Facebook     61.23%

The 28 sites we monitored received a total of 31,888 visitors signed into a Facebook account and 50,347 who were not.

Overall, this quantity of traffic is a lot lower than the other sites we measured. The most likely explanation for this is that instead of only having to load an image to trigger a response, it has to load a whole separate page, which makes the process a lot slower.

Google Plus Stats:

Users signed into Google Plus          5.40%
Users signed out of Google Plus     94.60%

The 28 sites we monitored received 5,800 visitors who were signed into a Google Plus account, and 101,589 visitors who were not.

Twitter Stats:

Users signed into Twitter.com          5.42%
Users signed out of Twitter.com     94.58%

The 28 sites we monitored received 5,330 visitors who were signed into Twitter.com and 101,585 visitors who were not.

B2B Stats:

B2B users socially signed in           13.97%
B2B users not socially signed in     86.03%

The B2B sites we monitored received a total of 7,861 users who were signed into a social account, and 48,395 users who were not signed into a social account.

B2C Stats:

B2C users socially signed in           14.63%
B2C users not socially signed in     85.37%  

The B2C sites we monitored received a total of 35,157 users who were signed into a social account, and 205,126 users who were not signed into a social account.

Overall Stats:

Avg. users signed into a social account           14.50%
Avg. users not signed into a social account     85.50%

Overall, 43,018 of the users we were able to track were signed into a social account, and 253,521 were not.

Some Additional Data:

Pinterest Stats:

Users signed into Pinterest           0.065%
Users not signed into Pinterest     99.93%

The 28 sites we monitored received a total of 682 visits who were signed into a Pinterest account, and 103,494 user who were not.

However, we decided NOT to include this information in our overall statistics because of how relatively new Pinterest is, and because it is still in such an early, formative stage, its influence might be limited in comparison to Google Plus, Twitter and Facebook.

Google Stats:

Users signed into Google            15.69%
Users not signed into Google      84.31%

The 28 sites we monitored received a total of 16,971 users who were signed into a Google account, and 91,125 who were not.

Although we're aware that Google is not a social network, we thought it would be beneficial to analyse it. At the time of implementing the experiment, the impact of the (not provided) keyword was not yet fully known as SSL had not yet been released as default in the UK. Since SSL is now default, it’s great to compare the stats that we found compared to the amount of (not provided) keywords to see whether or not a site is receiving a % close to the average. It’ll also be valuable to continue to monitor the Google sign in stats due to upcoming change to FireFox where all users using the updated FireFox will be using the SSL protocol as default. We can then compare how much of an impact the change to FireFox has had. You can find our analysis of this in the next section. Again, we have NOT included these statistics in our overall findings.

About the Experiment:

We began tracking all the social networks in our experiment (Facebook, Google, Google Plus, Twitter, Pinterest) on the February 22^nd, and we took our data after recording just under one month. We are hoping to do a follow-up when we have recorded a larger data sample and also to include LinkedIn and MySpace in the results (you can put the lack of LinkedIn data down to an unfortunate oversight on our part).

As is often the case with experiments conducted online, it is nearly impossible to create the vacuum necessary to conduct a test that will be completely unaffected by environmental or other external factors. As such, I'll disclose a couple of things that could potentially skew the results:

• Our experiments relied on JavaScript, so if a user views the site with JS disabled, they won't be tracked.
• If a user is using multiple browsers (signed into Facebook on Chrome, but visiting the site in Firefox) they will be tracked incorrectly.
• If a user is signed in using software such as Tweetdeck or Hootsuite, they will not be tracked.
• Server-side latency at one of the social networks could mean that the user will visit the website and leave before the response code triggers anything in analytics.
• The way browsers request the response codes might cause a user's visit to be logged, but they could leave before the JavaScript code triggers a response code. This could cause some discrepancies in the numbers.

What does it all mean?

Google

You always have to be careful when analysing statistics, because as the saying goes – statistics are like putty in a child's hand; they are what you make of them. However, there are some interesting patterns that we feel deserve commenting on.

The first relates to the quantity of users signed into a Google account. Obviously this now has repercussions for SEOs, inbound marketers and website owners around the world, as logged in traffic from referred from Google will no longer pass on keyword data beyond the ambiguous "not provided" epithet…which, as we all have to grudgingly accept, is useless.

Our data shows that 15.69% of traffic coming into our websites was signed into a Google account, which is a significant amount more than the 'single-digit percentages' that Matt Cutts speculated when Google first announced they were rerouting all logged in users to the SSL version of search. A number of other webmasters have already commented on the portion of traffic coming to their sites via the "(not provided)" keyword is already far exceeding the 10% threshold that Cutts suggested. The more worrying thing is that this number is only going to increase as more and more people sign up for Google accounts along with the upcoming change to FireFox which will soon add Google SSL as default.

There is also an 10% disparity between the quantity of users signed into Google Plus (5% signed in), and those signed into a peripheral Google account (YouTube, Gmail etc.). This would seem to suggest that although many people signed up for a G+ account when the network was first launched, there aren't a lot of active accounts, or at least not as active as Facebook. However, as I mentioned earlier, we do expect this number to keep growing, meaning that personalisation is going to filter into a lot more of the SERPs, particularly when SPYW is rolled out across all Google local domains.

Facebook & Twitter

The volume of visitors coming to sites while signed into a Facebook account far exceeded those who visited while signed into a Twitter account. This could be linked to the popularity of software such as Hootsuite and Tweetdeck for viewing Twitter, but there are still three specific examples I would like to focus on.

There were three sites we monitored that generated a far higher volume of socially active traffic across all the networks we were monitoring. I'm not able to name them, but they were a Sushi business based in London, a site where young aspiring actors and models could look for work, and a website that provides stylish laptop bags for women. All of these websites were aimed at a younger audience (20-30 approx.), and their social media presence was heavily emphasised on their website.

There was one clear pattern between the social media marketing strategies for each of these sites – Twitter was the hub, and there was lots of cross-promotion used to highlight their presence on the other social media networks, mainly Pinterest and Facebook. For example, the sushi website was using Twitter to promote new items on their menu, and they would add a link to a photo of the new item, which was hosted on their Facebook profile. Therefore, Twitter followers are made aware of their presence on Facebook and the brand is increasing their social outreach. The site selling the laptop bags also used a similar tactic, but instead of promoting their Facebook account via Twitter, they shared new images on Pinterest, and encouraged users to follow them there. Again, the intention is to increase outreach through the use of Twitter a central hub for social media activity.

B2B & B2C Websites

We found this to be one of the most interesting aspects of our survey. We predicted before the test that B2C sites would attract a higher volume of socially active traffic, particularly as we weren't monitoring LinkedIn users.

However, the difference was not as vast as we had expected, with B2C websites attracting only 1% more socially engaged traffic than B2B sites. Perhaps this indicates that even businesses whose target market is other businesses are also seeing value in the use of social media marketing. It is also interesting to note that B2B targeted websites received the highest portion of their socially engaged traffic from users who were signed into a Google account.

Conclusion

Overall, we found that 14.5% of the traffic that was coming into the sites we were monitoring was active on one of the social networks we tracked during the experiment. In isolation this might not seem like a particularly significant amount, however, if you think consider that there is an excellent opportunity to encourage 14.5% of your traffic to engage socially with you, there is huge scope to build the social media outreach of your brand. As well as having an impact on your brand visibility, this 14.5% of web traffic could potentially provide a huge benefit to SEO performance directly.

Here are a few tips we have come up with to help maximise the potential that website visitors will socially engage with a website. We recommend that webmasters use their discretion if and when they choose to implement these changes, as there are potential privacy issues:

• You could potentially only show social buttons to users signed in with the relevant social account. So, for example, if a user isn't signed into Pintrest, don't show a Pinterest button. This tactic allows you to tailor the website content more accurately to individual visitors, and potentially increase the likelihood that they might engage socially.
  However, this has several intrinsic limitations. If users visit your site, but they just aren't signed into a social account, you might miss an opportunity to create engagement. Also, visitors might not be aware of all the social media networks where your brand exists.
• You could pre-populate email fields with '@gmail.com' if the user enters your site while signed into Gmail. This makes the process of subscribing to a newsletter more efficient for the user, and could improve conversion rate.
• Bearing in mind that our study revealed nearly 15% of traffic is signed into one kind of social account, as I have said, there is a lot of potential to promote social engagement. Rather than simply displaying social buttons, try and make them more prominent, and featuring an appropriate call-to-action.
• You could use open-graph Meta data to get permanent links from Facebook and make your site appear and function almost like a sub/external Fan Page. This would be particularly effective if you notice that a lot of your traffic is active on Facebook, again encouraging engagement.
• It would even be possible to slightly alter your content based on the social networks the visitors are signed in on. For example, we know that Pinterest generally tends to attract more female users. So, you might decide to slightly tweak the images you show or your content based on whether the visitor is signed into Pinterest.

So, those are our findings, but now it's time to open things up to the floor – what do you think the significance of our finding are, and how are you optimising your websites to promote social media engagement? We'd really love to hear your thoughts!