Why the First Month of Social Network Usage is Crucial

Why the First Month of Social Network Usage is Crucial

Introduction

Success on social networks like Twitter depends on how many people you interact with. The number of followers is a simple measure of one’s status within the network. Having many followers indicates that your opinion is valued by others. Who you chose to follow (your followings) determines how much value you can achieve from the network. Too few followings, and information is limited to a trickle; too many followings, and you can experience information overload.

One question that arises is “what is the rate over time that a person acquires followers and followings?” This can give an indication of how quickly the network is growing, and provide insight for marketers on when to acquire new followers. In this blog post, I will examine in detail the time dynamics of followers and followings for the typical Twitter user.

Methodology

I used Conditional Independence Coupling[1] to create a sample of 36,435,499 tweets from 51,266 authors between November 1, 2011 and October 31, 2013. For each unique author in the sample, I collected the number of followers and followings, as well as the number of days since they joined Twitter. I grouped the data according to the month the user joined Twitter, and then analyzed the number of followers and followings within that group.

Who’s a Typical Twitter User?

First, I needed to determine who constitutes a typical Twitter user. As a first guess, I took the mean of followers and followings (Figure 1). The problem with the mean is that there is a lot of monthly variability in the followers and followings. This results in quite a few outliers. Merely taking the average does not get at the numbers for the typical Twitter user.

followers_mean

Figure 1: Mean Growth of Followers and Followings

Next, I used the median (figure 2). The median is a representation of the number of followers and followings for which exactly half of the authors are above and half are below. Using the median can reduce variability in data where there are large outliers. In this case, the resulting data is much more uniform.

followers_median

Figure 2: Median Growth of Followers and Followings

Analysis

Looking at Figure 2, the first thing to notice is that people acquire followers and followings at roughly the same rate. The actual numbers are 2.39 ± 0.13 followers / month and 2.47 ± 0.16 followings / month. This nearly identical number is consistent with a model where a typical Twitter user follows those who are following them. This leads to the question whether there is a correlation between number of followers and number of followings. It turns out that the link is weak at best. The next thing to notice is the number of followers/followings that are acquired in the first month (month 0 in the figures). Within the first month of signing up, the typical Twitter user will follow 120 ± 6 people and acquire 68 ± 5 followers. This suggests that people select whom to follow within the first month. After that window, people acquire followings more slowly.

Conclusions

What does this mean for brand marketing? First, you will have more success acquiring followers if you target people within their first month on Twitter. After users have established their followings, it is harder to convince them to follow you. Second, following more people will not automatically boost your followers. While the data is consistent with the typical user following those who follow them, there is only a weak correlation that this can lead to large numbers of followers.

If your company could benefit from analysis of the growth of your followers, or you would like lists of new Twitter users interested in your market, contact us for more information.

Featured “Smile in Crowd” Image courtesy of Alex Berger View CC License

Footnotes    (↵ returns to text)

  1. K. White, J. Li, N. Japkowicz, “Sampling Online Social Networks Using Coupling From the Past,” 2012 IEEE 12th International Conference on Data Mining Workshop on Data Mining in Networks, pp. 266-272

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.