Khoros Atlas Logo

The True Marketing Power of Facebook: Sociology Perspective

Lithium Alumni (Retired) Lithium Alumni (Retired)
Lithium Alumni (Retired)

Dr Michael WuMichael Wu, Ph.D. is 927iC9C1FD6224627807Lithium's Principal Scientist of Analytics, digging into the complex dynamics of social interaction and group behavior in online communities and social networks.


Michael was voted a 2010 Influential Leader by CRM Magazine for his work on predictive social analytics and its application to Social CRM.He's a regular blogger on the Lithosphere's Building Community blog and previously wrote in the Analytic Science blog. You can follow him on Twitter at mich8elwu.



Happy New Year, and welcome back. I hope you all had a wonderful holiday. This post is the sequel to my previous post The Social Dynamics of Facebook Fan Pages, which investigates an often overlooked weakness of FB. Previously, we explored the social dynamics that govern how people behave on the Facebook (FB) platform and particularly on FB fan pages. Due to the attention economy and the conflicting social spheres on FB, we actually arrived at a rather counterintuitive result. That being – the reason that a FB fan page (which is a community) is not engaging for a business is precisely because the FB platform (which is a social network) is too good at facilitating engagement among the strong ties. I must emphasize that this is not a technology problem. It is an inherent problem in how humans behave, because people naturally focus their attention on stronger ties and tends to ignore weaker one during any active engagement.


So what does this means, especially to brands and enterprises? It means that FB is superb at maintaining relationships (those that already have a relatively strong tie with you), but not very good at developing new relationships (those that have relatively weak ties with you). This shouldn’t be too much of a surprise to the cyber anthropologists here, because the social anthropological role of social networks is to maintain pre-existing relationships. If you already have existing relationships (assuming that they are good relationships) with your customers, then FB would be an excellent medium to engage these customers. However, if you don’t have existing relationships with your customers, then FB is not ideal for building customer relationships.


Now that you know about FB’s drawbacks, we’ll discuss its major strength today. There is no doubt that FB, as a platform, provides many benefits to individuals. Otherwise, there wouldn’t over 500 million active users on it. But today, we will focus on the business use of FB for enterprises. Since we will use the social principles we’ve learned, I recommend reviewing the following posts before diving into today’s discussion.

  1. Figuring Out the Relationship Puzzle
  2. The Relativity and Economics of Relationship
  3. Where is the New Dunbar Limit?
  4. Can Social Technologies Increase Our Dunbar Limit?
  5. Virtual vs. in Real Life: The Value of Relationship Perspective


Efficient Routing (Finding the Shortest Path) in Social Networks

To fully appreciate the true power of social networks, such as FB, I must give you a little historical context. So bare with me a bit.


I’m sure most of you have heard of network concepts like six degrees of separation and small world effect. In a nutshell, these mean that even though most people in the world are not neighbors of one another, they can be reached by anyone with small number of steps through a chain of friends. And this chain of friends often requires no more than six people. Note that there are certainly chains that are much longer, but the shortest and most efficient path, between any two persons is usually no more than six steps. So between Yu (the green bubble in figure 1 conveniently labeled Yu) and pretty much anyone on the earth (the red bubble labeled Al), there are only about 5 persons in between (the target, Al, is the sixth person) connecting the both of you (figure 1).




Although we know that there is a shortest path, with no more than six steps, the more challenging problem is, can we actually find these short paths?


Knowing this short path exists is of great theoretical importance, but to make this knowledge practical, we must be able to find them efficiently. So the big question is can Yu route a message to Al efficiently through the social network? That is, if we ask Yu (a random person in the world) to send a message to Al (another random person in the world), with the restriction that Yu can only send the message to her friends (people she knew at a personal level), and likewise for everyone else, how many steps will it take for the message to reach Al?


Clearly, if people continue to route Yu’s message through the social network, it will eventually reach Al. But how many steps will it take? Will we be able to find the efficient path (about six steps), or will we find paths that take many more than six steps?


To understand the challenge of this problem, let me point out an interesting observation: Within this shortest path, it is almost certain that you only know one person (two if you are in the middle of the path). As depicted in figure 1, Yu only knows Sue, but doesn’t know Ray, Kim, Jen, Ed, or Al (the target, which can be anyone on this planet). Between you and a fisherman in the Island of Kiribati (you probably don’t know where or even heard of Kiribati), there are probably only 5 people connecting the both of you. But how do you know who, among all your friends, will bring you one step closer to that Kiribati fisherman? In other words, how can Yu find that shortest path to Al without knowing anyone or anything beyond the first step?


If you think about this seriously, you will realize that this is not a trivial problem, because Yu doesn’t even know the fact that Sue and Ray are friends. Yu probably doesn’t even know who Ray is. So how can Yu know who to send the message to among all her other friends in the first place? And how can the network collectively find the shortest path to Al?


Navigability of Our Social Network

This problem has baffled many scientists, and consequently many experiments were carried out to investigate this. The most well known of these, is the Small World Experiment conducted by Stanley Milgram in 1967. The surprising and repeatable result of this experiment is that, we CAN collectively find these short paths, and it really takes only about six steps to route a message to anyone in this world. This property of our social network is known as navigability.


As opposed to the small world property, which only claims the existence of short paths (average no more than 6 steps) between any two random persons, the navigability property claims that our social network can collectively find these short paths efficiently. Therefore, FB is a very efficient medium for spreading pertinent information, if it is a close enough approximation to the true social network of our offline world. In business term, it means that FB is great for driving awareness and creating interest.


Let’s revisit the problem of how you would route a message to that Kiribati fisherman (your target)? You will need to look among your friends and find the one person who has some traits that might connect him/her to that fisherman. For example, you may realize that Kiribati is an island nation in central/south Pacific. However, you might not know anyone living in central Pacific now. So you might send the message to a friend who used to live in Hawaii, since that’s an island in the northern Pacific, and this friend may know people who live in Kiribati. Or you may remember that one of your childhood friends’ family owns a seafood restaurant, so his family may know someone in the fishing industry who can eventually bring the message closer to your target.


In most cases the connections will be very weak and extremely indirect, because we don’t have much information about the target. All we know is a foreign address, a name that we don’t recognize, and his occupation. Moreover, the target is a random person that has no share history, common interest, or any relationship to your life. Therefore, trying to identify these weak and indirect connections to a random person will require you to know your friends pretty well. For example, you would have to know something about your friends’ connections, family, work history, past residences etc. Otherwise, you might not be able to find the best person as one step of the entire routing process.


So knowing our friends deeply is clearly valuable. The fact that we knew our friends well is what enabled the rapid delivery of relevant information on social networks. Because we know what is relevant to our friends, their likes and dislikes, their wants and needs, we tend to pass on information that are useful and filter out those that are irrelevant with respect to our personal network. Moreover, people inherently trust their friends more (see Figuring Out the Relationship Puzzle), so on average, people tend to pay more attention to information from their friends and pass on more information from friends.


Realistic vs. Unrealistic Social Network

I must emphasize that this is a property of our social network in the offline world, and it is not specific to FB. You probably have heard of the adage that “gossips travel fast.” The reason is because it is traveling on our social network from one person to another person. If you’ve worked with a social network, you probably notice how rapidly news spread when there are people who want to spread it.


The reason that FB is an efficient medium for driving awareness and an effective medium for creating interests is because it is a pretty good approximation of our offline social network. FB tries very hard to capture and reflect the real relationships between real people. That is FB tries to ensure that you are who you say you are, and your friends on FB are indeed your friends in the offline world. In contrast, although Twitter has networking capability built into its core platform, it makes no effort to reproduce our offline social network. Consequently, many relationships on Twitter are not real. As a consequence there is much more noise on Twitter, and the information passed around on Twitter is not as relevant or trustworthy as that on FB.


Having More Fake Friends is Bad for You and Everyone

So clearly our social network (and FB) supports efficient routing of information. The deeper question is why, and how are we able to do it? Turns out the reason has to do with the structure of our social network. Despite the sheer size of FB (having 500M users and growing), structurally it consists of many tiny overlapping networks. Recall our discussion on the Dunbar limit? On average, each person’s personal social network (a.k.a their egocentric network) is only about 148 according to Prof. Dunbar. In fact, average users on FB only have about 130 friends. So FB is actually a collection of 500M tiny little networks of roughly 130 people that overlaps significantly.


Now, you may recall that there is tradeoff between how well you know your friends (tie strength) and how many friends you have. The more friends you have, the less you can know about each one (see Where is the New Dunbar Limit?). This means you ability to find those weak and indirect connections will be degraded, when you have more friends. Let’s do a simplistic analysis of what could happen if people increase the size of their egocentric network just a little bit, so their ability to identify those weak and indirect connections are only degraded by 10%. What is the effect of this on our network’s navigability (i.e. collective ability to find the short paths)?


If figure 1 depicts the shortest path from Yu to Al, then Yu knowing 10% less about his friends could translate to Yu only have 90% chance of routing the message to the right person (i.e. Sue). What’s the big deal? The big deal is that this is happening for everyone. So Sue (or whoever Yu routes the message to) will also have 90% chance of routing it to the correct next person along the path (i.e. Ray). What happens is the error compounds each step along the way. So Yu will only have 81% chance of getting the message to Ray. Continuing this calculation, the probability of the message reaching Al would be (90%) x (90%) x (90%) x (90%) x (90%) x (90%) = 53%. This is barely above random chance (which is 50%).



FB has definitely created much value for their end users, but what is its greatest value for businesses? This question is rarely analyzed from a sociology and relationship perspective. I hope this article gives you a new perspective on what you might already know.

  1. Although there is only six degrees of separation between any two random persons in this world, finding that chain of six people (the shortest path) requires collective efforts from many people and it’s not an easy problem.
  2. Yet, it has been demonstrated that our social network is able to collectively find those short paths. This property is called navigability.
  3. This means that our natural social network is an efficient medium for spreading relevant information. For business, it is great for driving awareness and creating interests.
  4. FB’s true strength is in how well it mimics our real social network in the offline world. Without this, the information on FB would be just as noisy as those on Twitter.
  5. Based on a simplistic analysis with lots of assumptions, if everyone gets more friends and knows 10% less about each friend, our social network’s navigability could be wiped out almost completely.

The small size of our egocentric network not only enabled us to know our friends better, it also keeps information that is only relevant to a few well contained within one’s egocentric network. On the other hand, if the information is relevant to a broader audience, then the navigability and the small world property of our social network would enable each person to spread it rapidly to a large audience. This balance between relevance and reach is what gives FB to power to drive interest and awareness so effectively. By establishing a presence on FB via a fan page, brands can automatically take advantage of this power.


Alright, this post explored quite a few sophisticated network concepts. Next time I will try to wrap up our sociological discussion about FB and its business use cases. As usual, I welcome kudos (by clicking on that star symbol to the lower right), comments (by responding below), suggestions, critiques, and all forms of discussions. See you next time!



About the Author
Dr. Michael Wu was the Chief Scientist at Lithium Technologies from 2008 until 2018, where he applied data-driven methodologies to investigate and understand the social web. Michael developed many predictive social analytics with actionable insights. His R&D work won him the recognition as a 2010 Influential Leader by CRM Magazine. His insights are made accessible through “The Science of Social,” and “The Science of Social 2”—two easy-reading e-books for business audience. Prior to industry, Michael received his Ph.D. from UC Berkeley’s Biophysics program, where he also received his triple major undergraduate degree in Applied Math, Physics, and Molecular & Cell Biology.
Matt Kammerait
Not applicable

Very interesting points Michael. Always a pleasure to read.


An added variable plays into the mix of "finding the shortest" path as well - the active algorithmic filtering of streams by Facebook. A little known fact to many people is that their top news is being actively filtered by Facebook to encourage interaction - this means many of their "friends" and the companies they follow will never have an update appear in this (crucial, most important) of feeds. So, they'd have to play an active role of searching for the content, which, as we know, most users won't do (unless a pre-existing relationship exists).


This relates back to Milgram in an interesting way as well - in Six Degress - participants were asked to actively use their network for greatest reach - for most individuals this isn't an "every day" or common behavior. It's rare that we try to reach more than a step or two into our network. So this closes the door for likely interaction (and "new" friend generation) even further.


Fascinating stuff! Keep it coming.


Also, I can't get your Kudos button to work in either Chrome or IE... unfortunately - so text-based kudos!

Lithium Alumni (Retired) Lithium Alumni (Retired)
Lithium Alumni (Retired)

Hello Matt,

Welcome back. Thank you for stopping by and being the first to comment on my 2011 post.

Good point. Yes. Facebook (FB) has an active filtering algorithm that determines what shows up on each user's news feed. It's called Edge Rank Algorithm. So not all updates from a fan pages (or even friends) show up on every fan's feed. In fact, FB’s Edge Rank favors strong ties through an affinity score, which gives more weight to entities (friends or fan pages) with whom you have sent many message, check their profile frequently etc. So again, if you already have a strong relationship with a particular fan, he/she will see more of your fan page updates. Otherwise, fan page updates are likely to be filtered out.

This add to the thesis of my previous post that FB is not great at building new relationships, but great for maintaining pre-existing ones. Because compare to the strong ties, people's relationship with brands are usually much weaker. Therefore, update from fan pages/brands has much greater probability being filtered out. Unless the fans actively search and view a fan page frequently, which means the relationship already exist.

Concerning Milgram’s Small World experiment: In real life, word of mouth (WOM) diffusion is a passive process on our social network. Such passive diffusion can only propagate through our first degree connections (our friends, 1st degree connections, people who connect to us directly). We can’t really have passive diffusion through people that are more than 1 step away, because we don’t know them.


One interpretation of the Milgram experiment is to mimic this diffusion process. That is why they limit participant to only send the message to friends (1st degree connections). And the key finding is that even only through 1st degree connections, a message can reach any desired target rapidly (within 6 steps), as long as the information is relevant. So as long as the message is relevant on FB, they will get propagated provided they don't get filter out by Edge Rank along the way.

With Kudoes, you need to register and be part of the Lithosphere community in order to give kudos. We'd love to have you as a part of Lithosphere. Either ways, thanks for the comment, and see you again next time.


Frequent Commentator
Frequent Commentator
Hi Michael,
An interesting read once again.

Agree with you on the power of navigability. The idea of generating awareness by finding folks "closer to" your target desitination by modeling the appropriate attributes (in the example you gave that would be 'living in Hawaii' to target a fisherman in Kiribati) is not new. In the database marketing world, the look-alike models had a similar objective -- but I think the beauty of Facebook is that this "idea" is repeatedly applied by each recipient within the same context and so the information reach and propagation is tremendous while keeping the relevance in check.

In the traditional database marketing world after the first set of targets (based on modeling) recieves the information, any future propagation is mostly done by WOM (except in certain cases like email campaigns) and so the likelihood of the information travelling far is pretty slim.

You also bring a very important point into play. To fully leverage the power of Facebook, it is not enough to just create a fan page or focus solely on getting more fans. The power is in profiling your fan base so that when the need arises to create an awareness campaign or create interest in something you already know the seed population to target. This will as you said maximize the liklihood of optimizing reach and relevance.

And lastly, I was thinking that while the targeted approach is pretty powerful, FB is also a good pond to fish in. In other words, instead of consciously targeting someone (e.g Yu targeting Sue) one could create the right bait and leave it out there. Of course, this is not a good or recommended methodology for everyone but if your fanbase is pretty strong and diverse/you have a good FB network then chances are that a 'quality/relevant' fan will pick up the information and propagate it to the right audience. What do you think?

Thanks for a good read.
Lithium Alumni (Retired) Lithium Alumni (Retired)
Lithium Alumni (Retired)

Hello Ned,

Thanks again for the comment. And thanks for pointing out the interesting connection to the look-alike model in DB marketing. I actually didn’t know about that.

Concerning profiling, in most practical situation, marketers only can target a certain seed population for WOM. The rest is WOM diffusion. Marketer in general do not need to think about routing the message through the social network to reach someone in particular. That is, if they know Sue is a potential customer, they will just directly contact Sue. The problem is they don’t know who are the potential customers and they have no way of contacting them. So they have to rely on WOM diffusion to reach these unknown potential customers.

This situation is like a routing a relevant message to someone you don’t know (the Milgram experiment). Because, each step along the shortest path, a user is simply passing the message along to his friend who might have some interest in the content of the message. The only difference on FB  is that each person is actively pulling contents that he finds somewhat interesting from his network, rather than pushing them to someone in his network. The difference of push vs. pull in this case do not alter the result. The message still diffuses through the social network through each person’s 1st degree connections. This two problem is in fact mathematically equivalent.

However, because FB fan page is not very good a building relationships, it makes profiling difficult. Brands never get to know their fans that well enough to profile them. That is why the next best thing to do is, as you said, to create the right bait and leave it there and some fans will find it interesting/relevant and will pick it up and share it. (If you know who these relevant fans are, you would have target them as seeds in the first place.) Likewise, there is no guarantee that our friends will continue to share what we shared. We can only hope that one of our friend find this interesting enough to share it to his network. And this is how WOM propagate on a social network.

Anyway, I hope I didn’t create more confusion than clarification. Thanks again for the conversation.

Frequent Commentator
Frequent Commentator

Hi Michael,
First really interesting read in 2011!


I realize FB as clustered social space with theoretically high level of navigability. But in practice?

Problem is to know key seed persons|communities for best concret (thematically) routing paths without perfect  knowing (today at least) about existing interest (relevance) networks that regulate a distribution and recognition of spreaded content. Good contemnt is a part of deal. Another part is optimization of diffusion. And I don't know some perfect mechanics or tools to manage diffusion process beyond 1 steps of distribution. But they most come?..


And FB community? Must we build at all one more closed fan community (through pro strong relationships managment within) - and so make navigability of relevant content around it community more difficult? It question is actual if we will understand main goal in driving wide awareness. But does it matters?


Thanks for great sociological investigations about balance between relevance and reach!

I am proud to have similar investigations about social conditions of information routing as it happens with virus messages through WOM mechanics (my article with mentions of you as "pro-social" author).

Lithium Alumni (Retired) Lithium Alumni (Retired)
Lithium Alumni (Retired)

Hello Andrei,


Thanks for the comment.


Finding the seeding population in a WOM program is a similar (nearly equivalent) problem to that of finding influencers. And I’ve already written extensively about the finding influencers (I even compiled a chapter describing the model and algorithm for finding influencers). The beauty of the Milgram’s result is that the structure of our real (offline) social network support efficient routing. That is, even if we don’t know the routing path, information can still find its way to relevant target rapidly through our social networks such as Facebook, because it is a good approximation to our offline social network.


I am not sure if you can optimize diffusion beyond choosing a good initial seed population. Because there is nothing you can do beyond the first step. If you know who the 2nd steps would be and you want the information to go there, then you could have target them in the first place. But then they become your first step. So it really comes down to selecting a set of good seeds for the initial targetig. Some rule of thumbs are:


     1. Don’t choose seeds that have too much overlap in their network (so their effects don’t overlap)

     2. Use big seeds (so there are many sources of propagation)

     3. Focus on influencers (since they give you the most bang for your buck)


Concerning building community. I think that my next post will address your question. So please stay tuned!


Finally thank you for your interest. And thank you for citing our whitepaper (I have to use Google Translate to help since I cannot read Russian). Thanks again for coming back to comment. See you next time.



Frequent Commentator
Frequent Commentator

Hi Michel!

Thanks for comment!

I agree with your rules for best start in seeding. It is current situation, that we can not manage diffusion process as distribution actions of peope (they are a good will of them). 

But after seeding happens nurturing! Does smm manager have a look at diffusion process after managing initial seeds. We have social graph (not enough open) and social networks (not enough analyzed fron SNA and interest point of view). Is it possible to manage additional seeding at breaking points - where some networks overlap but do not diffuse (maybe it is some given influencer or some "blocker")?


It requires a lot of info and big and micro view at the same time - but is it a development direction for this industry?

I hope to be not so imaginative and groundless 🙂


Do you have some another cases for WOM campaigns/modelling through influencer seeding?

Lithium Alumni (Retired) Lithium Alumni (Retired)
Lithium Alumni (Retired)

Hello Andrei,


Thanks for coming back and keeping the conversation going.


It is possible to do some level of monitoring after seeding. That is what Social Media Monitoring or Listening Platforms do. If you seed the a fairly unique meme, you can track the propagation of such meme from your seeding sources. But still, if you identify a blocker or a part of the social graph where your initial seeds didn't reach, you can certainly redo the seeding again.


The FB social graph is open through their OpenGraph API. And that social graph is a good model of our real social network. However, it is true that we don't use Social Network Analysis (SNA) enough to analyze the data on FB's OpenGraph. There are not enough tools out there for doing large scale SNA.


This is not too imaginative. It is something that we are currently working on as we speak.


We currently do not have other white papers on WOM. If you are interested in my influencers' work, you can check out my chapter on influencers. Moreover, I will probably revisit the topic of influence again since that seem to be a topic that has confused a lot of people. So stay tuned for more discussions on influence and influencers.


Thanks again for the discussion. See you later.