Google Friend Rank or FriendRank to Rank Social Marketing Profiles

FriendRank, the new PageRank for Social Marketing Profiles

FriendRank is actually called “NETWORK NODE AD TARGETING” in the patent application but has already been dubbed FriendRank or FriendRank ala PageRank.

Filed on December 29, 2006, Google has been tweaking this Friend Rank algorithm for quite a while. Now public as of July 3, 2008.
Google describes this in the patent as:

1. A computer-implemented method for displaying advertisements to members of a network comprises identifying one or more communities of members, identifying one or more influencers in the one or more communities, and placing one or more advertisements at the profiles of one or more members in the identified one or more communities.

So in other words, Google is going to provide an advertising server, that will target advertising to all, some or just on the profile of the most highly ranked (by Google) social site user’s profiles. The most important part of this so far is that Google has….

Been doing this since before December 2006,

And That Google can currently and in the past, determine which social site users are “influencers” as Google in their own words says that they “rank” users of social sites.

This will depend on:

  • How active you are (submissions, votes, comments)
  • Your interests depending on your participation
  • How many friends you have and their interests
  • Who and How active your friends are and their submissions, votes, comments
  • What groups you AND your friends are in
  • Who else is in those groups, their rank and your friend status with other users

We have seen this many times in Digg’s algorithm, where the same group of Digg users constantly make posts popular and reach the front page. Case in point, September 26th, 2007 at 3:27 pm, Keven Rose’s profile shows that he has submitted 295 stories to Digg, but somehow, 298 of his submissions have reached the front page.

Why is this important to Bloggers and social bookmarking users like you and me?

Because one of two things was going to happen very quickly….

Google was going to dismiss all social bookmarking sites as major sources of spam since they are being over frakin run with junk posts. Spam posts just to create back links and traffic has been burying Digg especially, for a while.

OR…

Google was going to come up with a way to decide what posts were actually real submissions, from real users that posted the content because it interested themselves and they REALLY wanted to share it with their friends and participate in the network for more than their own gain.

Now with the public release of this patent, it is not the posts that are going to be judged, but the profiles of who submitted them.

Also, if a particular post is WAY outside the users normal interests, Google could even find specific posts irrelevant due to unusual content. Not probable, but possible.

Google also states that advertising revenues could be shared with powerful profile owners. How apt do you think other users are going to be to submit your blog posts if it could hurt their Google income?

Now, some of you are going to read this and say “This only applies to a Google advertising model” and you are right to say that. But if Google is able to rank social site profiles as being what they call “influencers” don’t you think Google will use FriendRank to also determine the validity of profiles as well?

You have to make the jump here and consider that if Google is serving advertising on these profiles then the content had darn well better be of value. If Google and their social algorithm has been out there for a year and a half they have tons of data to play with then don’t you think that they already know what profiles are there only to create traffic and backlinks?

Next, wouldn’t you think that a Google social bookmarking slap is coming to those profiles that contain nothing but links back to the same site?

Let’s step through the Google FriendRank patent one item at a time

The entire first section (known as the claims) of the Google NETWORK NODE AD TARGETING application # 20080162260 is taken verbatim from the US patent office. Noting is added or removed and the claims section appears here in it’s entirety, each line is highlighted in yellow. All other text is the interpretation of myself, Chris Lang.

2. The method of claim 1, further comprising receiving information pertaining to a plurality of members belonging to the network, wherein the information comprises content of profiles of the plurality of members and links between the plurality of members.

Google finds a social network that has clickable links that Google bot can follow.

This is Google talking about tracking and identifying each member or a social network and who their friends are, easily done when the links they speak of are HTML a tags (clickable links).

3. The method of claim 2, wherein identifying the one or more communities comprises grouping members based on the links established between the members of the network.

Whether you belong to a common group within a social site or that Google can see you as a social group because you have common friends.

4. The method of claim 3, wherein the link connects a first member with one or more members of the network.

Now Google takes the first social profile it finds but that profile must have outgoing links to other profiles. So, if you have no friends, you are going to be found as irrelevant. Why would a social profile have no friends? Because it is an automated spam profile with an email address that is never answered when friend requests come in.

Or it is a dead profile that has been abandoned.

5. The method of claim 3, wherein each link has a weight.

This is really important, that “each link has a weight”. So the more links you have can be counted as links coming into your social profile and hence more important to Google. Just like our sites are ranked, Google is also going to rank our social profiles by the number of links in the social site to each profile.

It does not matter if the link is nofollowed, Google can still follow it. That is why they say “each link has a weight.” They are taking nofollow out of the equation in FriendRank.

Important: “each link has weight” meaning that each link in your friends list is counted and has a weight. Each link has “a” weight, can’t say that that means each links has a weight of one, or each link has it’s own algo determined weight. I think the more powerful your friends are the more link weight they have to pass on to you.

6. The method of claim 2, further comprising identifying one or more common interests of the plurality of members of a community based on the content of the profiles of the plurality of members.

Now Google is going to try to create a “common interest” profile of a group of social site friends by looking at the keywords that are in each profile. If it is a social bookmarking site they have even more data to play with by spidering your submissions.

7. The method of claim 2, further comprising ranking the plurality of members belonging to each of the one or more communities based on the links between the members of each community.

I read “ranking the plurality of members” and “based on the links between the members” as ranking each individual member within a social group by the number of friends they have and their level of participation. Also I would say that if your friends are the bargain basement type with few of their own friends and low participation then you are going to be ranked much lower.

8. The method of claim 7, wherein the influencer is the member with the highest rank.

We are now identifying the leader of the group and finding #1. Again I believe this is about participation and positive reaction. In social bookmarking this one indicator here would be a high positive vote percentage compared to other users and friends. Plus participation, comments, votes on others submissions and other interactions that are unique to particular networks.

9. The method of claim 2, wherein the network comprises an online social network.

THINK about the fact that the patent keeps talking about a “server.” This would allow Google to serve advertising on even walled networks like MySpace. But only if there was a Google server given access to the network.

Look at Ning, you can set your Ning group to be visible only to users who can log in. However Googlebot has always indexed Ning sites even though browser access is limited to logged in users. So Google could still identify “influencers” and server advertising based on their profiles.

10. The method of claim 9, wherein the profile comprises one or more web pages stored on a server hosting the online social network.

The profiles of users must be crawlable by Google, a web page comprised of HTML.

11. The method of claim 1, wherein a member of the network belongs to one or more of the plurality of communities.

The social site profile must be part of groups or communities within the parent site. Joining groups is a sign of being legit and social. This allows more targeting of the interests of each user profile and can bypass social profiles no matter how well constructed to hide your own content for your own gain.

If you are a member of groups or communities then you are going to be visiting these groups pages and viewing advertising targeted to that group.

Also by determining the interests of the most powerful users of a social site group they can server advertising group wide to all who participate.

12. The method of claim 1, wherein the one or more advertisements are placed solely on the profiles of the one or more influencers.

Google can also decide to only place advertising on the profile of the “influencer.”

13. An advertisement server for displaying advertisements to members of a network, the advertisement server comprising: a community identifier configured to identify one or more communities, and identify one or more influencers in the one or more communities;and an advertisement inventory configured to store advertisements to be displayed on the profiles of one or more members of the identified communities.

You have to have a market that can be monetized and an advertiser that is willing to spend advertising dollars. They identify the market, then serve the ads stored to be served under certain criteria. Think AdWords and keywords.

One reason advertising has failed so heavily on social networks is that it is not laser targeted to the interests of the users. Again, think AdWords on social profiles that is so focused that can’t help but click. Remember that Google get’s paid on clicks, not views.

14. The advertisement server of claim 13, further comprising a network inventory configured to store the information pertaining to the plurality of members belonging to the network, wherein the information comprises content of profiles of the plurality of members and links between the plurality of members.

This is Google spidering and storing (“content of profiles”) what it finds in a social network on it’s own server so then can further analyze for demographics and keywords.

15. The advertisement server of claim 14, wherein the community identifier is further configured to group members based on the links established between the members of the network.

The actual algorithm going to work here, determining the grouping of members to target advertising to.

16. The advertisement server of claim 15, wherein the link connects a first member with one or more members of the network.

Now Google starts over again…

17. The advertisement server of claim 15, wherein each link has a weight.

Links are counted…..

18. The advertisement server of claim 14, wherein the community identifier is further configured to identify one or more common interests of the plurality of members of a community based on the content of the profiles of the plurality of members.

Further demographic interest based analysis….

19. The advertisement server of claim 14, wherein the community identifier is further configured to rank the plurality of members belonging to each of the one or more communities based on the links between the members of each community.

Identify each users interests….

20. The advertisement server of claim 19, wherein the influencer is the member with the highest rank.

Google looks for another #1 power user(s).

21. The advertisement server of claim 14, wherein the network comprises an online social network.

In a social network…..

22. The advertisement server of claim 21, wherein the profile comprises one or more web pages stored on a server hosting the online social network.

Where Google can spider it because it is a web site…..

23. The advertisement server of claim 13, wherein a member of the network belongs to one or more of the plurality of communities.

In social communities….

24. The advertisement server of claim 13, wherein the one or more advertisements are placed solely on the profiles of the one or more influencers.

Highly targeted advertising on the “influencers” profiles.

25. An advertisement server for displaying advertisements to members of a network, the advertisement server comprising:a means for identifying one or more communities, and identifying one or more influencers in the one or more communities; and a means for storing advertisements to be displayed on the profiles of one or more members of the identified communities.

The advertising engines goes to work again…

26. The advertisement server of claim 25, further comprising a means for storing the information pertaining to the plurality of members belonging to the network, wherein the information comprises content of profiles of the plurality of members and links between the plurality of members.

Then Google stores this new set of profiles….

27. The advertisement server of claim 26, wherein the means for identifying one or more communities and identifying one or more influencers in the one or more communities is further configured to group members based on the links established between the members of the network.

Ranks the new found group by the same algo criteria….

28. The advertisement server of claim 27, wherein the link connects a first member with one or more members of the network.

You must have frineds to be found valid, removing spam profiles and dead profiles.

29. The advertisement server of claim 27, wherein each link has a weight.

Counts links……

30. The advertisement server of claim 26, wherein the means for identifying one or more communities and identifying one or more influencers in the one or more communities is further configured to identify one or more common interests of the plurality of members of a community based on the content of the profiles of the plurality of members.

Demographic algo…..

31. The advertisement server of claim 26, wherein the means for identifying one or more communities and identifying one or more influencers in the one or more communities is further configured to rank the plurality of members belonging to each of the one or more communities based on the links between the members of each community.

Counts each member in a group thereby ranking the group….

32. The advertisement server of claim 31, wherein the influencer is the member with the highest rank.

Indentify #1…

33. The advertisement server of claim 26, wherein the network comprises an online social network.

Online…..

34. The advertisement server of claim 33, wherein the profile comprises one or more web pages stored on a server hosting the online social network.

On a web server….

35. The advertisement server of claim 25, wherein a member of the network belongs to one or more of the plurality of communities.

The member is active in groups or communities….

36. The advertisement server of claim 25, wherein the one or more advertisements are placed solely on the profiles of the one or more influencers.

Hignly targeted ads are served on just the #1 profile….

Lather, rinse, repeat!

The bottom line

Google also looks for overlap where group B may be interested in advertising that is served on group A because there are many members of group A in group B.

This FriendRank algorithm could be used for so many other things than serving advertising, ranking news, ranking blogs, forums and social profiles and removing social profiles from the index if they are considered to be spam oriented.

This will be the first in a two week long set of articles on the new Google FrinedRank patent, there is a lot more to come.

13 Comments

  1. Posted July 15, 2008 at 10:27 am | Permalink

    If this ends up helping to devalue to hundreds of thousands of fake user accounts used for the sole purpose of spamming social sites – I’m all for it!

    People will have to come up with ways then of putting value into their profiles and, again, I’m all for that too.

  2. Posted July 15, 2008 at 10:58 am | Permalink

    @Jack Humphrey, I agree, what I really find interesting is that the patent application date is Dec, 2006.

    I expect this to be part of the Google slap that usually comes in October. Not to mention that Google has been updating the Websearch algo about 10 new points a day for the last few weeks.

    @All, So if you are sticking in your search positions and not sinking then you MAY be able to expect to bypass the fall slapdown.

    Always good to hear from you Jack. = Chris Lang

  3. Posted July 15, 2008 at 11:06 am | Permalink

    Do you think that eventually this could result in the devaluing of links submitted and voted on by “worthless” profiles, translating to an assumption that the links they submit and vote on are also “worthless” and resulting in a reduction of page rank for sites that seem to be heavily bookmarked by people that google deems to be “worthless”?

    In other words, if they use this for more than targeting ads, this has the potential of affecting the PR of a lot of sites.

  4. Posted July 15, 2008 at 11:24 am | Permalink

    @app, to answer your questions, yes, Yes and YES. I have always felt that the Google slapdown on social bookmarking is coming, this is just the start.

  5. Posted July 15, 2008 at 11:40 am | Permalink

    I usually get involved in things right before the slapdown. YIKES.
    Who or what determines what and who is worthwhile? I think I need to read this a few more times to digest it.

    Chris needs a nickname like, watchdog Chris, or something.

    Thanks for the post.

  6. Posted July 15, 2008 at 12:32 pm | Permalink

    Chris, this is a great analysis & I really look forward to the upcoming additional posts on this topic!

  7. Posted July 15, 2008 at 12:34 pm | Permalink

    My reading of this as that you want a social profile with a lot of incoming links (friends). And you’re better off if you can cross link your profile across many different social networks.

    Frankly, I don’t see this as a whole lot different than just calculating the internal link juice on these sites while ignoring nofollow. If a user has only a few incoming links and their stories don’t get many votes, the total link juice will be pretty small anyway. If you’ve been spinning out fake accounts to single vote stories, you’re in trouble. But if you’ve been diligently building a strong profile with lots of friends and your stories get lots of votes, you’ll be about the same. The big difference is that the big G is going to try to change which ads those power users see.

  8. Posted July 15, 2008 at 12:34 pm | Permalink

    Chris,
    Another great post, I have been watching all the things I do on social networks and I also watch the search engines. I have been honest on all the social sites I am involved in. I am glad that I found you and Jack before I buried myself.
    @all before you leave subscribe to this feed and take in all you can.
    Thanks again Chris for all you do for all of us.
    Keep it flowing:-)
    Rheyanna

  9. Posted July 15, 2008 at 1:35 pm | Permalink

    Now this was really interesting. I am looking forward to following you through the logic of all this to see where the yellow brick road will lead us all. Groupie, sure why not. Let’s do it.

    Stephen

  10. Posted July 15, 2008 at 2:39 pm | Permalink

    Chris,

    You lost me on some of the details in this post, but I think I get the ‘jist’ of it. The bottom line is: gaming the system will not work forever and the only real sure bet is to provide stellar content that “real folks” want to Digg is the only way to win in the long run. Right…?

  11. Posted July 15, 2008 at 4:22 pm | Permalink

    @Mathew Hunt, to be honest this article took me two days and I had to chop up the pieces of the pie here to get it myself, one piece at a time, and there is much more in the patent application.

    But I agree, content that is not just yours and focused is the way I see it too. Digg, though is another ballgame all together. Just be careful of any kind of Internet marketing type articles on Digg. They don’t like us much there.

  12. Posted July 16, 2008 at 2:42 am | Permalink

    I have to say it is interesting. I just have to keep doing what I do for my business and clients and believe that as long as good content and honest effort are represented I can weather any Google storm coming. Thanks for the forecast!

  13. Posted July 16, 2008 at 2:52 am | Permalink

    @Marianne, you are exactly right, as long as you are not doing something that you think could get you in trouble then you are on the right path. Also, since Google has been tweaking and testing this since 2006 then if you are seeing good search result rankings and you have not taken a sudden nose dive in the last few weeks. then you are probably going to be fine.

2 Trackbacks

  1. […] we can see from the FriendRank patent that was made public July 3rd, Google already has been planning to rank social profiles with a […]

  2. […] long will it be before Google starts using their FriendRank patent to rank our profiles, deliver advertising on them and start showing a FriendRank bar in Google […]