A Marketer’s Introduction to Social Network Analysis
For many, Social Networks begin and end with Social Media, but network analysis is about much more than tweeting. As the tools improve, Social Network Analysis will become as important to marketing as target group analysis was a generation ago.
Understanding networks will help us bring order to the complexity and chaos in the world around us. From how information spreads to managing sales and finding the information we need, networks can help us run our businesses better.
Fortunately, all you really need to know about networks, social or otherwise, you can learn on your next airline flight.
The Importance of Hubs
On your next trip, once you have paid strict attention to the safety procedures, take a look at the airline magazine in the seat pocket in front of you. If you open it to the back, you will find a map of showing all the airports your airline flies to with paths between them designating routes.
If you take some time and start counting, you will see a pattern develop. The biggest airport will have roughly twice as many routes leading to it as the second biggest airport, three times as many as the third biggest airport and so on. The bigger the airline you are flying on, the closer these proportions will hold.
Mathematicians call this proportional scaling. They denote the relationship as 1/s and it holds for many naturally occurring phenomena, such as income and population distributions, amazon.com sales, ecological systems, etc. It has a distinctive shape that Chris Andersen made famous as the long tail.
Another point of interest is the overall proportions. With some more checking you will find that roughly 20% of the airports account for about 80% of the routes. Over a century ago, Vilfredo Pareto recognized this pattern in economic data. Many managers know that this 80/20 rule also holds for their businesses as well.
You don’t have to look very deeply to see that much of our world works like an airline map.
Anti-Hub Strategy – The Case for Efficient Network Paths
When we travel, we want to choose the most efficient network path. Our airline travel is just a means to an end so we want to spend as little time and money as we can on it. The airlines share some of our concerns but not all.
Some airlines choose a hub strategy. They have strong presences at major airports such as London Heathrow, Chicago O’Hare and others. Through investing in these hubs they are betting their ability to take us to all the places we want to go will be valuable enough to us for them to make a nice profit.
Others, like Southwest and Ryan Air, choose a different strategy altogether. For the most part, they bypass hubs and can’t take us everywhere we want to go. However, for the places they can take us they can do so very cheaply.
The Benefits of a Robust Network
We have the best situation of all, because we know exactly where we want to go. If a well traveled route takes us to a preferred destination, chances are that we will find a very cheap flight. If not, we can go through hubs and still get to where we want to go. In either case, travel agents have software with fairly simple algorithms that will help us find the most efficient network path.
The richer and more diverse the network, the bigger the hubs will be and also the more ways around them. Well traveled routes will give us a variety of options, while more out of the way places will inevitably cost more.
We see the same thing in business. For many products, we can just market to the masses very cheaply and the customer network will do the rest. We can even help the process along by encouraging people to forward our message through social media. (See How Ideas Spread)
However, some products are niche and require us to spend more by targeting specific network paths through trade and event marketing.
If you page back in your airline magazine a bit, you will find the “feature well” with some longer articles for your reading enjoyment. If you have some time, you can make a list of the words in each article and how many times they appear and you will find that the mathematics of networks apply here too.
Once you have finished, you might notice the same thing that George Zipf did in 1932. The distributions of words follows the same 1/s law that holds for airline routes and amazon.com sales. In effect, language also follows network laws.
However, texts are a bit different because they are evaluated holistically. It is natural to speak of good writers and bad writers and word distributions can give us a clue to discern between the two. Better writers usually have wider vocabularies and therefore have wider word distributions in their texts. The wider the choice of paths, the more ways to gain efficiency of expression.
While it is common to think of written texts this way, it is also true of anything that follows network rules. For instance, we like Amazon because it gives us a lot of efficient network paths. Just like a great writer has the vocabulary to find the perfect word, we can usually find just what we’re looking for on Amazon.
To see how we can apply this simple rule to any business, here’s an example of two different businesses with approximately the same revenues.
As you can see, the sales follow the familiar “long tail” pattern. A small amount of customers spend a lot, and a lot of customers spend a little.
Due to a algebraic quirk, we can evaluate networks mathematically quite easily. With a fairly simple technique called a log-log plot, we can transform our network from a hard to work with “long tail” to a simple downward sloping linear graph that we used in high school.
Moreover, just like in high school we can evaluate the curve by taking the rise over the run. The smaller the value of our slope, the wider the revenue distribution, the more complex interactions and the more robust the network is.
If we apply the same technique to our two businesses, we can see that our two “long tails” are actually quite different.
Here, we can clearly see that while business A is more reliant on big customers, Business B has a more diverse customer base, even though they have roughly the same sales and number of clients. Moreover the slope give us a good numerical way to rank diversity: the flatter the slope the more diverse the revenue sources are .
While the math might seem complicated, it’s actually pretty easy to set up an excel template to calculate everything.
How Networks Grow
Lets take another look at word distributions. We’re not born with a full vocabulary, we have to learn each word.
When we were young, our parents were delighted to hear our first word. Eventually, our vocabulary became richer and we learned different categories of words. Simple words we use a lot and it would be tough to communicate without them. Other words are used much less but can take on greater importance depending on context.
It really doesn’t matter where we start our network journey because we can always add new connections. Einstein was a late talker – in German – but later became highly respected for the things he said in English. Facebook didn’t start at a party school, but at Harvard.
There hasn’t been a marketing strategy invented that’s a match for happy customers and word of mouth. Brand networks grow through the the people who are devoted to them. Given enough passion, anybody can be an “Influential.”
Network Marketing Applications
In business, we will find three common applications:
Robustness: By measuring the diversity of a network, we can get a good idea how strong it is. A business that is based on a few hubs is more vulnerable to external shocks.
Broadcast: In order to sell our wares and make money, we need to get our message out. The connections in networks help us to do that. Moreover, just like a worldwide tour can start in any city and a sesquipedalian’s vocabulary can start with any word we have a variety of ways to broadcast.
In effect, we want to find the most efficient network path. Hubs are useful, but we don’t want to overpay. The real issue best strategy will allow us to cover the network most efficiently, which requires us to not only consider hubs but also paths around them.
Targeting: As mentioned above, networks are rich in information contained both in objects and in the connections between them. Whenever we need information, chances are that we have access to a network that contains it. Targeting is the converse of broadcast because we want to narrow the paths rather than maximize them. In effect, it is a search.
The key with a network search is to not make it too broad by trying to search an entire network and not make it too narrow by using too many parameters. Experimentally it has been shown that two or three parameters is enough to perform an efficient search. Any more than that and we will most probably exclude valuable network paths that could lead us to our destination efficiently. (See Social Search)
This issue will come to the fore as digital marketing becomes more pervasive. Just because we can exclude people doesn’t mean we should. We’ll need to be careful not to over-target.
Using Network Theory to Help Form a Marketing Strategy
The more varied our customer base, the more stable our business will be and the less vulnerable to market shocks. If we find that our customer base is lagging behind our competitors, that’s a good sign that we need to go after new segments.
Another key point is that we want to find the most efficient way to reach our goals. That might not always be the most targeted or the most direct. We can often initiate more network paths cheaper through a mass audience.
Just like when we go to a travel agent, there are lots of things to confuse us. Fancy travel posters of exotic locations, frequent flyer deals and other gimmicks shouldn’t change the simple fact that we want to go somewhere quickly, cheaply and with some comfort.
Fortunately, just like with our travel plans, we can navigate the networks that are crucial to business with a little information, some common sense and a good idea of where we want to go.