Given all the juicy excitement around the many dramas over the last while (Bing stealing from Google, The smackdown trifecta of JC Penney, Forbes and Overstock or the two Google updates) one little goody seemed to fly virtually un-commented under the radar; the Google social search update.
I’ve been covering social and real-time search pretty much since the beginning and at most times have generally poo poo’d it as far as being an area of SEO that we really need to be overly interested in. By it’s temporal nature alone, it has left much to be desired. While the social graph was interesting, (Google has an API for this) there was no clear sign on how it would have made a larger impact on search behaviour.
Tucked away at the bottom of the results page or in a side bar vertical (real time and ‘latest’), serious impact seemed minimal at best.
Fast Forward, Getting Personal
This latest change has actually done something far more than I had ever anticipated. It furthered Google’s over-all goals and has added a new universal element to the SERPs with far reaching impact. One of the stated areas of extreme importance at Google over the last 5 or more years was; personalization. Not only does a tighter granular level of personalization offer users (hopefully) greater relevance, it also can help engineers to handle spam better (concept being that you can’t spam yourself). All things they sorely desire lately.
Of course to get there, many obstacles remained, the least of which being how they could glean strong signals through implicit feedback (see the; Final Word on Bounce Rates as a Ranking Signal). Oh and of course this result came up while looking for that post;
Anyway, the interesting part was that over the years search engineers had some difficulty finding strong signals from implicit data. This has made deeper personalization a problem. Another of course is the sheer processing required. I doubt even the Caffeine update took Google personalization to a granular person-by-person level (currently it’s calculated on categorization segments of similar users).
So how does one find new levels of personalization when faced with these challenges? Look elsewhere. In this case, people’s social graphs.
More Than Just a Few Faces
Ok, great… all very interesting but what does it have to do with these changes and my apparent interest? Because this isn’t about just seeing little avatars in your search results. It is actually adding pages (to page one) and re-ranking others. Much like we saw with Universal Search elements over the last few years, there is a NEW way to get to page one (results). Or if you’re already there, to move up the food chain. If THAT isn’t SEO, then I don’t know what is.
Let us consider this search for [SEO radio]
And logged in
Not only does it re-rank the results in this case but it is also annotating it with mentions from my social circle. Depending on the query space and the person, this is also a HUGE enticement to increase your SERP CTR (click-through rate). Given how many articles we still see each week on meta-descriptions and CTR, it’s interesting that this still hasn’t gotten people talking (for that element alone).
This is certainly an interesting end-around from Google to add new levels of personalization and it most certainly is something we’re likely to see more of, not less, moving forward. Those that recognize this early, will likely benefit the most.
So, What Are They Doing?
The next stop along this journey is to get a sense of what exactly they might be up to as far as mining connections and information. As far back as late 2009 yours truly has been writing about how social personalization might effect us. Even then, Google’s Marissa Mayer had been talking about a ‘social influenced PageRank‘ type concept…. and even further back I’d looked into the so-called ‘FriendRank’ patents from Google. If you’re serious about learning more, be sure to read those.
We certainly know they are looking at the social graph in some sense to sort out who is going to affect this personalization. For that I’d have a look closer at the social graph API. To get a sense of what that can do, I’d suggest looking at this demo application. Mine looks like this;
The connections themselves come from;
- Google services ““ Google contacts, Buzz, FriendConnect, Google Reader (people or websites you follow)
- Websites or blogs ““ this can be done by identifying them in your Google profile
- Social Services ““ Twitter, FriendFeed, YouTube, Digg, Quora and more.
And of course many of these can be added/connected via your Google profile. This is where, in particular, you can add website connections that may fall outside of the ‘social graph’ such as myself and fellow authors here at SNC.
The deeper the profile, the more connections there are likely to be. Which of course means that early on there will likely be a higher connection rate in markets where geeks roam. Even so, these are early days and I’d start to get a feel for things and how they work.
Are they actually looking at the tiered connections to see whom I have more connections with? Are they looking at whom in your social circle you’re interacting with the most? That part seems unclear.
Profiles Are On the Grow
Interestingly, shortly after the changes to social search, the Google profiles also had a facelift. If you haven’t looked lately, here’s mine.
While the changes aren’t drastic, they are certainly getting a little more social in flavour and I get that will continue to be a strong element of their social plans moving forward. I would suggest that you go and play around with yours if you haven’t lately. Once again, becoming more intimate with the evolution can only help.
You can even check out the new search interface by adding the parameters; &tbs=prfl:1 ““ to any Google search. Here’s an example;
Interestingly, it also brings up those you are connected to before it does strangers (for a generic first name search). This further shows how the connections in the social graph are being made.
Where Is Social Search Headed?
Of course we also want to get a sense of where all of this might be headed right? As marketers it is important to establish the final destination to get a sense of how much time we want to invest in this directions for ourselves and our clients.
From the social profiling patents we get a sense of it along these lines;
Open Profile (user identification) – Uses identification and scoring/categorization analysis of a user profiles
- demographic data
- interest/topic categorization
- media data
- group associations
- influencer score
Custodian profile (content relational) – Looks at inferences between the viewer of a web page and the person that created, or manages it.
- user profile data
- user acquaintance data
- user group data
- user media data
- user options data
- and other user data
Relationships and Topics – For looking at common relationships among users and related topical categorizations (and performance metrics).
- Common groups
- Common behaviour
- Similar membership in groups
- Similar profile data
- Common acquaintances
- Other similarities
If we add to this the other elements of personalization; search history, surfing history and geo-localization, we can certainly get a sense of what potential they’re seeing here and how data classifications might play over the coming years. I don’t believe we’re as deep as we can get, (the above signals are a huge laundry list) but social is such a fast evolving area that the search engineers themselves are likely struggling to create and test new signals as social morphs and grows.
What Does it Mean to SEOs?
And so, my weary wandering web warrior, we must consider exactly what this all means. Once again I have always taken the stance that very little of what Google was doing with the social graph really meant that much to search optimizers. Sure, we do pay some attention to social, there wasn’t enough there to really consider it deeply. That seems to be changing.
Now that it’s starting to become not only a standard part of the results (placement and rankings) it also fits well with Google’s desire to find a deeper personalization in search. With all of the problems with spam, relevance and (lack of) explicit signals, I can see the attraction here. There are more than a few other social elements that Google is said to have in the works, it seems this will only continue to evolve.
But what can we do now? I really haven’t had enough time to formalize a strategy but some early elements one can do include;
- Make sure the talking heads have profiles (PR folks, social reps, staff etc..)
- If using Gmail or other Google products; re-consider the strategy and expand outreach.
- Align social channels ““ go beyond Twitter and Facebook. Add other services to your profile.
- Align social connections ““ try and cross reference and follow the same people on as many services as possible
- RSS is Back ““ if people read/follow you in Google reader, it is considered a connection.
- Promote ““ I mean duh… promoting yours encourages others to develop theirs, which is in your favour (yes, I’m working on it for SNC).
- Share wisely ““ as always, the value of your recommendation (in the SERP) depends on the trust. Ensure social channels are being metered.
Those are truly some just off the top of my head. I must be candid and say that I’ve not been as excited about a SERP change since the big picture of Universal Search (strategy) knocked me over the head. As always some follow up research, playing with implementation and measurements will be required, but early on this seems to be a move we can count on not only sticking around, but evolving more in the near future.
Consider if you will that other personalization method we mentioned; local. People are entities. Google likes entities (more on this soon). So are businesses. How about a review and recommendation engine deeper into Google local? Consider the statement;
“While profiles work well for individuals, we’ll continue to work on new ways for businesses to engage with their customers, so stay tuned for updates. “ – Google Social Web Blog
Anyway… I could literally go on for hours on this. While everyone was playing pin the tail on the big brand, I’ve been musing endlessly on where this one lands. I for one am (finally) excited.
Oh and… be sure to hook up with me on the various social channels in my Google profile, and I shall see YOU in my SERPs real soon!
UPDATE; UK Style
It is worth noting that this update, much like the Panda Farmer Affair, hasn’t hit all the various international data centers. I have talked to some UK folks that aren’t seeing it yet. So, we shall keep an eye on it.
On another note, there are avatars beside Twitter accounts showing up in the UK and not here, (in .com and .ca). Here’s a ew examples from Kev Strong and Barry Adams;
While not entirely related, it does show other interests in mining the social graph and even the use of avatars in the actual SERPs. Anywathing else we come across, I shall certainly update this post.
I am psyched to see more social signals showing up in GOOG. Starts to make me look like I knew what I was talking about when I wrote this a year and a half ago 🙂 http://searchengineland.com/are-you-ready-for-google-local-social-search-28873
It’s certainly been a bit of a process that’s for sure. They do seem to be making some interesting moves lately though. The comment on the social blog about ‘businesses’ also make one wonder the direction this all might take (thinking reviews here as well). Oh crap… SEO is dead right? As long as all the competitors believe it, tis ai’ght with me. hehe
So Dave, with your research have you seen any information about a cross over from non personalization? What I am getting at here is this post:
You can see there that the engines both say that social media user that share links and their perceived influence can attribute to rankings. When they say that are they just talking about personalized results that are already influenced through the Social Graph API? Or regular SERPs?
Anyways, great post and summary.
Joe, I have definitely seen Facebook activity have a serious impact on non-personalized BING SERPs.
It’s possible Joe, but as with any implicit signal when it affects the greater index, the higher the risk and chance of noise. This is why I found this an interesting end around to the problem. Keep in mind personalization as we know it isn’t granular – it’s related user type groupings. Thus there is already some inherent risk of noise I’d imagine.
Now percieved influence on rankings, could mean via social graph. Meaning in the personalized environment, WHO from my connections shows up and who doesn’t? Who has the power to re-rank my results on a given topic? That approach seems sensible. I have to believe the huge risk here is the lack of tight management tools for the feature.
From here I am going to keep reading, testing, watching and searching to see what it does. I’ve spent almost 3 years going ‘meh’ at social search, it’s finally getting interesting lol.
Great stuff as usual Dave. I have seen social become increasingly important over the last couple of years. I think the increased influence of social signals opens up so many interesting opportunities and I am looking forward to the progression.
BTW I dig that you searched for Mark just so you could get a picture of me in this article. Very smart idea.
This is a really great article Dave. I hadn’t thought about this before because I was thinking of personalization from the point of view of an end consumer. But, for branding or spreading your influence in your niche (and all the attendant extra links/mentions that may entail), personalization is going to have a much bigger effect on those that are more connected to the internet for business reasons.
This is huge. I have to go think for a while now. Thanks so much.
Dave – Thanks for the thought provoking article. As Google improves their capability to evaluate connections, it will be interesting to note if they use this to downgrade the value of links for web rings, tribes, and close social media connections.
Enjoyed the post, Dave and Eddie here with Topsy Labs.
We’ve been playing around with “social search” for a while, which for us means being able to harvest and search for links shared within a users social graph. The challenge is data sparsity, meaning most consumers have relatively small social graphs (at least right now), so you get really sparse results searching with one degree of separation. You can open up to 2+ degrees, but this has a pretty significant degrading effect on content quality.
I think an ideal approach is to blend social graph and influence-based results together. This provides the consumer with links shared among their social graph, plus links shared from people that have high influence related to the keyword term being queried.
We just had a whitepaper published here http://searchengineland.com/using-influence-to-tune-signal-to-noise-on-the-social-web-66602 that speaks to how influence is used to rank search results & increase content quality.
Love to get your deep thinking into this stuff.
I like your post here really informative. I was just wondering about it’s impact. I think I’ve to look into more details about it so I would understand it further better since I’m not totally aware of that. Thanks a lot. Love your post each time.
Comments are closed.