Do Streamer Demographics Impact Twitch Channel Growth?

Is the rate of success of Twitch’s finest streamers predetermined by characteristics like skin color, sex, and relationship status?

by S4B0T4G3FIRE | February 19, 2021, 8:00 AM EST

Abstract

Society is organized in such a way that many people are destined to fail or succeed based on factors that are out of their control, regardless of how much education they have or how much hard work they put in. This study explores the same societal structure as it applies to the success of Twitch streamers. A sample of 300 streamers was selected, with an equal representation of male, female, single, in-a-relationship, white, and of-color streamers. The results are the sum of 18 months’ worth of observational data collection. Additionally, the number of channels that no longer exist, underwent name changes, used bots to illegally gain followers, and received “Affiliate” or “Partner” status was also recorded for discussion. The trends identified are rather conclusive seem rather conclusive.

Seeking Specific Demographics

For as long as Twitch has existed as a platform, its viewers have had “free range” over where they went, who they watched, and (most importantly) who they followed. Although the concept of “free range” gives the impression that users’ navigation of the platform is entirely random, it is theorized that their navigation is actually skewed toward streamers of particular characteristics. According to a survey that has been taking place over the past year, users actually look for very specific details before clicking on a new channel to watch. The results of that survey are as follow:

• Profile Pictures (2.6%)
• View Count (15.4%)
• Titles (38.5%)
• Thumbnails (43.6%)

The percentage beside each criterium is the percentage of respondents who consider that particular characteristic most before clicking on a channel. As you can see, “thumbnails,” which display the most accurate representation of a streamer’s facial features, gestures, and other expressions, is the most important criterium to the average user navigating the platform. This means there is reason to believe that viewers’ free range is biased toward a particular appearance and, quite possibly, a particular sex and/or race as well.

The Study

Now that we are assured of users’ biased hunts for content-creators, we can delve into whether or not these initial notions are manifested in the channel growth of said content-creators. Are some streamers predestined to grow quickly or slowly due to their demographics and the so-called “viewer bias” toward those demographics? This is where an 18-month-long study is carefully analyzed. The parameters of this study are as follow:

Sample Selection

In order to represent the majority of the streaming platform, the sample selected is comprised of channels from Twitch’s top ten categories (where the majority of streamers and viewers are concentrated) as of June 25th, 2019. These categories are listed (in no particular order) below:

• Fortnite
• League of Legends
• Just Chatting
• Grand Theft Auto V
• Dota 2
• World of Warcraft
• Counter-Strike: Global Offensive
• Overwatch
• Hearthstone
• Minecraft

Thirty (30) channels from each of these categories were strategically collected to be organized by:

• Male
• Female
• Single
• In A Relationship
• White
• Of-Color

(To be certain that the demographics were accurate, biographies were read, and streamers were respectfully asked questions under zero pressure of answering.)

**Sample Size = 30 channels * 10 categories = 300 channels**

Baseline Observations

The day before the commencement of this study, channel statuses (“None,” “Affiliate,” and “Partner”) and follower counts were recorded for each of the 300 channels.

Data Collection

On random days throughout the 18-month period, each of the 300 channels’ follower counts was recorded and averaged within each demographic subsection. In other words, the follower counts of female streamers were averaged, as were the follower counts of male streamers, single streamers, etc., during each data collection session.

Final Observations after 18 Months

Three (3) final observations were made once the 18-month period concluded.

• The status of each channel to see how many channels improved from “None” to “Affiliate,” and from “Affiliate” to “Partner.”
• The number of channels that either changed their name or no longer exist (ie. quit or were banned).
• The number of channels that experienced botting, which is when a certain number of accounts are sent to a channel to “Follow” that channel illegitimately/illegally.

NOTE: To avoid any confusion as to whether or not a large gain of followers was the result of botting or a host, Videos on Demand (VoDs) and the streamers’ social media were examined for evidence of hosts or botting. This means that the number of botted channels presented in this article is as accurate as possible.

Results: Bias’ Role in Determining Channel Growth

How did the “viewer bias” toward specific characteristics affect the channel growth of certain demographics? Let us take a look at the abundance of data below.

NOTE: Once any of the 300 channels became “illegitimate” (by quitting, changing their name, getting banned, or using bots), their follower count was halted, making it as though they never gained another follower for the remainder of the 18 months. This was done to nullify their growth after becoming illegitimate, without totally nullifying their previous legitimate growth.

Figure 1: Follower Growth of Channels Run by In-A-Relationship and Single Streamers

Figure 1 shows how follower growth varies among channels run by streamers who are single or in a relationship. Although the random selection of streamers resulted in a 53-follower “advantage” for Single streamers, In-A-Relationship streamers were able to close the gap and perhaps even gain the “upper hand” by the end of the study. Correlation does not equal causation, but let us discuss some possible reasons why this data may or may not have met expectations. Perhaps some of the following thoughts came to mind:

• In-A-Relationship streamers would gain followers more rapidly because they are more “understanding,” “personable,” and “committed.”
• Single streamers should have gained followers more rapidly because they are more “available” and “desirable.”
• Relationship status is not often a topic of discussion between streamers and viewers. Many streamers, in fact, prefer (rightfully so) to keep their personal life separate from their streaming career. This means that, in most cases, a single streamer would be indistinguishable from a streamer who is in a relationship, which explains why the growth rates are so similar.

Remember, these are nothing more than assumptions that should be avoided at all costs. The reasons why the results took the route they did are at the mercy of far too many factors to keep track of, but it is good to start thinking about the “why.” Let us move on to the next graph.

Figure 2: Follower Growth of Channels Run by Of-Color and White Streamers

Figure 2 shows how follower growth varies among channels run by streamers who are of color or white. This time, the random selection of streamers resulted in a 49-follower “advantage” for White streamers, which, over the 18-month period, was widened to a 130-follower gap. This is an increase of 165%. Once again, correlation does not equal causation, but let us consider why this data may or may not have met expectations. Perhaps some of the following thoughts came to mind:

• The majority of Twitch users are white.
• Since the majority of Twitch users are white, it may be easier to find (and therefore “Follow”) white streamers due to their abundance on the platform.

Although the first point is factual, according to Twitch’s own demographic data, the second one is an assumption that should be taken lightly until more data is collected and analyzed. Let us move on to the final graph.

Figure 3: Follower Growth of Channels Run by Female and Male Streamers

Figure 3 shows how follower growth varies among channels run by streamers who are female or male. Notice how the random selection of streamers resulted in a mere 22-follower “advantage” for Female streamers (the smallest lead among all three data comparisons), yet that gap is rapidly widened to 178 followers by the end of the study. This is an increase of over 700%! Once again, correlation does not equal causation, but a drastic change like this is still worthy of discussion. Does this data meet your expectations? Perhaps some of the following thoughts came to mind:

• Since the majority of Twitch users are male, perhaps they are seeking out female streamers to watch and support as though Twitch is a dating website.
• Historically speaking, women have generally been the more caring and empathetic sex, so perhaps Twitch’s predominantly male userbase is looking for someone who is relatable and/or openly cares about his/her audience.

Once again, these are nothing more than assumptions. Obviously, times are much different now, so it is not fair to assume any of the above. Whatever the case may be, any definitive proof as to why the gap widened as much as it did is still hidden behind further research.

Results: The Illegitimate Channels

Keep in mind that not all of the success shown in Figure 1, Figure 2, and Figure 3 was deserved. In this article, “illegitimate” is used to describe the channels that underwent some sort of legal (name change) or illegal (botting or suspension/ban) change and were not tracked for the entire 18-month period.

In the following table, you can see the percentage of illegitimate channels in each of the six demographics.

Figure 4: Quantity of Illegitimate Channels

• 4 out of 6 demographics shared very similar “dropout” numbers of approximately 10 out of 50 (20% of each sample). These demographics were “In-A-Relationship,” “Single,” “Female,” and “Male.” So, there is not much variation in dropout rates when sex and relationship status are the deciding traits.
• 2 out of 6 demographics shared heightened “dropout” numbers of 15 out of 50 (30%) and 18 out of 50 (36%), respectively. These demographics were “Of-Color” and “White.” Also, channels run by White streamers experienced the most illegal botting for followers, and channels run by Of-Color streamers experienced the most name changes, quits, or bans. So, when the sample is arranged by race, dropout rates are 10% to 16% higher than when organized by any other trait.

What could be the reason for this? One could guess that this shows how diversity is beneficial to the platform and its streamers and communities (which is obviously true), but your guess is as good as anybody else’s.

Results: The Journey to Affiliate and Partner

So, the streamers did their best to gain followers, but how many of them actually managed to upgrade their status from “None” to Affiliate” or from “Affiliate” to “Partner?” Let us consider the following pie charts to see how our streamers finished the 18 months (NOTE: Some illegitimate channels were still counted in the following charts in order to prove a point):

Figure 5: The Progress Made by Channels without Status

• 18 streamers (28%) either quit or changed their name in hopes of a fresh start.
• 30 channels (47%) remained stagnant.
• 16 channels (25%) reached Affiliate status.

Figure 6: The Progress Made by Affiliate Channels

• 36 channels (18%) either quit or changed their name in hopes of a fresh start.
• 167 channels (82%) remained stagnant.
• 0 channels (0%) reached Partner status.

Figure 7: The Progress Made by Partner Channels

• 1 channel (33%) either quit or changed their name.
• 2 channels (67%) remained stagnant, of course.

Additionally, the following events took place throughout the 18-month period:

• 4 channels without status gained Affiliate status after using bots to gain followers.
• 25 Affiliated channels maintained their Affiliate status after using bots to gain followers.
• 1 Affiliated channel gained Partner status after using bots to gain followers.

Although streamers are usually innocent of this terms of service violation (since bots are usually sent anonymously to channels), it probably feels pretty unfair to the rest of the consistent, hard-working streamers that a whopping 10% (30 channels) of the (close to 300) channels in this study jumped to a higher status via illegal means.

Conclusion

To summarize this entire article in as few words as possible, here are all of the results and takeaways, written in an easy-to-understand list. These correlations do not equal causation, so do not generalize them to the whole of the live-streaming industry.

• The thumbnail on a live stream is the most important factor to consider when trying to get people to click on a stream.
• There is no distinguishable difference between the follower growth of streamers who are single and streamers who are in a relationship.
• There is a small difference between the follower growth of streamers who are of-color and streamers who are white. The advantage appears to be that of White streamers, but not by a lot.
• There is a large difference between the follower growth of streamers who are female and streamers who are male. The advantage, within the scope of this study, is clearly had by Female streamers.
• When the sample is organized by race, dropout rates are 10% to 16% higher than when organized by sex or relationship status.
• Only 16 (25%) of 64 small channels became Affiliates after 18 months.
• 0 (0%) of 203 Affiliated channels became Partners after 18 months.
• 30 (10%) of nearly 300 channels jumped to a higher status (either Affiliate or Partner) via illegal means.

Thank you for reading this article! I really enjoy tracking data over time in order to answers questions that I or my readers may have about the way Twitch and other social media platforms work. To analyze 18 months’ worth of data in one article was very enlightening since it can provide some insight into the effects that demographics have on success. If you have any comments, questions, or suggestions for future studies, please feel free to comment below or on my social media at Twitch, Twitter, and Reddit.

Like this article? If so, I encourage you to check out some of my other articles here!