Social Networking

The Viral Expansion Loop

In most of the other articles in this series, I’ve talked about using search-engine marketing to inform the rest of your messaging, positioning, and targeting choices. But actually, if you’re the sort who’s inclined to really dig through data and do some math, then the information you garner from your social networking efforts can be even more valuable. The reason is the way that information spreads in a social network.

Now that’s a social network in action. If she told two friends, and the next day they told two friends, and the next day they each told two friends, and so on, then it would take one month (32 days, to be precise) for the entire world to know about the product. Of course, it doesn’t actually work out that everyone tells two friends. In reality, most people don’t tell anyone, and some people tell lots of people. The question is, can you figure out who is doing the telling?

Enter the viral coefficient. The viral coefficient is a term (as the name implies) from biology that is used to characterize the spread of a virus in a population. Now, perhaps a virus isn’t the most pleasant metaphor for your business, but math is math. From a social networking perspective, the viral coefficient measures the number of new members each new member brings. So, back to that TV commercial, word was spreading with a viral coefficient of 2. In reality, viral coefficients greater than 1 are extremely rare. But if you find one, you have a gold-mine on your hands.

Let’s use Facebook as an example. On Facebook, as a business, you can create a “page”. People can then become “fans” of that page. When they become a fan, three important things happen. First, you’re able to communicate with them: these are the people who will see the updates you’re making on Facebook. Second, the fact that they became a fan is shared with their friends, who then might also become fans of your page, and so on. Third, once someone is a fan, you are able to see the information they share about themselves on Facebook. What information they share is up to the individual, but you can usually see things like what city they’re in, what college or high-school they went to, and how old they are. In other words, basic demographic data. Over time, you can study the people who are becoming fans of your page, and learn with which group your message is best resonating.

So, say that you initially start a Facebook page for yourself and connect with all your friends and associates who are already on Facebook. You create a page for your business, and send a message to all of your “friends”, asking them to become fans of your page. Since these are all people who are 1st degree contacts with you, a large number of them are likely to do so. So, back to our paddle sports business example that we’ve been using throughout this series, let’s say our intrepid entrepreneur initially gets 50 of his friends to become fans of his page. Now, fast forward a quarter, and his page has 223 fans. As he went back and studied the data, he found that he could divide his initial audience into 3 groups with different viral coefficients for weekly growth: Group A was men between the ages of 25 and 45, and had 25 members at the start; Group B was women between the ages of 25 and 45 with 20 members at the start; and Group C with only 5 members at the start was men over the age of 55. Over time, the fan-base for his page looked like this:

Viral Growth of Page Fans in Facebook
Group A Group B Group C
Viral Coefficient 0.5 0.7 1.1
Initial Friends 25 20 5
Week 2 38 34 11
Week 3 44 44 17
Week 4 47 51 23
Week 5 48 55 31
Week 6 49 59 39
Week 7 50 61 47
Week 8 50 63 57
Week 9 50 64 68
Week 10 50 65 80
Week 11 50 65 93
Week 12 50 66 107

Now, from a total numbers perspective, this isn’t exactly exciting. Your number of Facebook Fans is a lot like the number of people who have signed up for your newsletter. Going from 50 to 223 in a quarter is pretty good growth, but 223 people isn’t exactly a giant target market. But that’s not the point: the point is how fast each of the sub-groups are growing. When you started your Facebook page, Group C represented a mere 10% of your fans. By the end of the quarter, they represented nearly half your names (and your total list had gotten more than four times bigger along the way). You can see what’s happening in the chart below:

Graph of viral growth curves for coefficients greater than 1, 1, and less than 1

When you have a viral coefficient of 1, you get steady growth. If your viral coefficient is less than 1 your growth curve flattens out, and in the rare case that it’s greater than 1 your growth is explosive! Now, that’s twice I’ve mentioned that viral coefficient greater than one are rare. In reality, if you find some way to market your product that results in a viral coefficient greater than one, then you can probably stop marketing for quite a while. If we continue the previous table through another quarter’s growth, look what happens:

Week 135066123

Viral Growth in Quarter 2
Group A Group B Group C
Week 14 50 66 140
Week 15 50 66 159
Week 16 50 66 180
Week 17 50 67 203
Week 18 50 67 228
Week 19 50 67 256
Week 20 50 67 286
Week 21 50 67 320
Week 22 50 67 357
Week 23 50 67 398
Week 24 50 67 442

As you can see, Group C has come to truly dominate your fans. By the end of quarter 3, they would have grown from 5 fans to almost 1,500 and would represent over 90% of your total population. And remember, that’s a viral coefficient of 1.1 and weekly growth. If the coefficient were higher, or the rate was faster, these numbers would quickly become unbelievable.

So, like I said, if you get a viral coefficient greater than 1, you’re probably done marketing for a while. But for the rest of us, what do we do with this data? We use it to inform our other marketing campaigns. Our paddle-sports store owner now knows that he gets a tremendous response from men over the age of 55. He can use that data to target all of his other marketing efforts.

Now, the one thing I haven’t discussed so far is how to actually calculate the viral coefficient. As I defined it, the viral coefficient is the number of people added to the group by each person. If you look at the data for week 2, then we see that Group A added 13 members to the original 25: (38-25)/25 = 0.5. You can do the math from groups B and C, and see that their viral coefficients between weeks 1 and 2 are, as stated 0.7 and 1.1 respectively.

Now, the truth of the matter is, the viral coefficient is normally used for projecting into the future. It’s used to forecast growth. When looking backwards, the truth can be harder to see. For example, if we continue to look at Group A, and calculate the viral coefficient between weeks 10 and 11, we get (50-50)/50 = 0. If we didn’t have the original starting point, we would think that this group has a viral coefficient of 0. But, it really does have a viral coefficient of 0.5, even in weeks 10 and 11 — it’s just that there aren’t enough members in the group to see it. If you just had the starting data, you might be tempted do do something like (50 – 25)/25 = 1 … which is not the right answer.

To see why that didn’t work, it’s important to remember that the viral coefficient is used for growth over time. By comparing week 10 to week 1, we fail to include the fact that the growth has already compounded 9 times in between. You can think of it in terms of your savings account: there’s a difference between APR and APY. If you put $1,000 into your savings account at the beginning of the year, and it has a 10% APR, you know you won’t have $1,100 at the end of the year, because your interest compounds monthly. In reality, you’ll have $1,104.71. The same principal applies with the viral coefficient — you have to know how many periods are involved in order to use it.

Conclusion

Well, there you have it. This brings my five-part series on introduction to on-line marketing to a close. I hope you both enjoyed it, and learned something from it. As always, if you have any question, please feel free to email them to me, or leave a comment here.

Cheers!

Chris