The Most Comprehensive Collection of Data-Supported Theories about How To Succeed on Twitch (TL;DR Edition)

Four years of deep research studies, data collection, and data analysis all culminating in one “how-to” guide to doing everything within a person’s control to succeed as a Twitch streamer.

by S4B0T4G3FIRE | March 19, 2022, 8:00 AM EDT

Introduction

There is no universal recipe for success, but there is data, and that is just as valuable. With data, you can visualize infinite cause-and-effect relationships between variables and predict how informed decisions will affect your future. Doing just a little bit of research now can help you plan out your childhood, so you arrive at a particular goal during adulthood. Having complete control over your future is arguably the most important thing in life.

For a teenage boy who wants to stream video games to thousands of viewers daily, what must he know? What must he do? What videogame(s) should he stream? Which days of the week should he stream on? How many hours per day should he stream for? Which hours of the day should he stream at? Is it worth the extra effort to stream on weekends? How should he go about mitigating harassment in chat to sustain a healthy, inclusive environment? With help from the data presented in this article, the teenage boy can answer all of these questions to give himself the best possible chance of achieving his dream. Your dream, perhaps. So, what could it hurt to place a bit of faith in some real data?

Getting Views

Part 1: Finding the Right Game

Threads and forums all over the internet will say that “You should stream whatever you find fun.” This may be highly encouraging to read, but it is of very little use to someone who dreams of climbing the ranks. While passion is crucial, it is silly to dive head-first into streaming on the notion that passion alone is a path to streaming success. There is much more to it.

What should I stream?

The “Top 20 Categories” are at the top for a reason. That is where the viewers are and where the money is. That is what gamers want to play and what viewers want to watch. That is where corporations want to dump all of their money for advertising, esports, and sponsorships. These categories are set up for success, and it would be a mistake for you not to take your share of that goldmine. So, before deciding to go all-in on your favorite game, consider things like:

• Games with esports
• Viewer-to-streamer ratio in certain categories
• Games that experience large, frequent, and/or exciting new updates
• Social media trends that boost the popularity of certain games

Figure 1: Raw Data from October 2021’s Most Streamed Categories

For October 2021, the criteria suggested that League of Legends, New World, Dota 2, Counter-Strike: Global Offensive, Valorant, and Call of Duty: Warzone were among the best categories for an aspiring Twitch broadcaster to stream. This changes from month to month, however. Besides, this is just a small factor in the algorithm. Many more questions need to be answered, and much more data needs to be factored in before a streamer can confidently embark on their journey.

Part 2: Finding the Right Stream Schedule

Once again, leave it to threads and forums to advise you to “stream whenever you have time,” but that simply will not cut it for the goal you have in mind. As it turns out, there are quite a few factors that influence this answer. For instance, the ratios of viewers to streamers change every day, and different categories experience fluctuations every day. This could boost some channels and hurt others. To help identify these fluctuations, an enormous sample of 11,000 English Twitch broadcasts from October 2021 has been collected (from TwitchTracker.com) and thoroughly analyzed.

Disclaimers:

• TwitchTracker was only referenced for a sample collection so trends could be identified.
• None of the sample data below was collected using “data scraping.” It was collected manually (ie. a human typing publicly-available data into a spreadsheet because it explicitly states on TwitchTracker’s website (as of the publication date of this article) that “Scraping is prohibited.”
• No data found on TwitchTracker is being redistributed anywhere in this article.
• This article is for non-commercial, educational purposes only.

Which days of the week should I stream on?

Sources say that the viewer-to-streamer ratio of the entire platform is worse on weekends than it is on weekdays due to a superfluous number of streamers, so the hypothesis for this segment is that “streamers’ views per hour will be lower on weekends.”

Figure 2: Average Views Per Hour according to Day of the Week (top 2 values per viewership category are highlighted in green; lowest value per viewership category is highlighted in red)

The 11,000 broadcasts yielded quite an inconclusive batch of day-of-week data, but there are still some important takeaways. Day of the week does appear to influence views. The data even goes on to disprove the hypothesis. At no point during the study did the lowest views per hour fall on a weekend. In fact, many of the highest views per hour fell on or around the weekends (Friday, Saturday, and Sunday). Monday broadcasts performed poorly, but Tuesday looked okay. A streamer can very easily do a month of trial runs on different days if they wish to confirm the best and worst days for them specifically, but it seems worthwhile for them to put in a bit more effort on the weekends.

To reiterate, this is just a small factor in the algorithm. There are still a few more factors that require consideration.

How many hours per day should I stream for?

Many sources say that “3–4 hours make for a good broadcast,” so that is the hypothesis for this next section. The data regarding stream duration looks pretty concrete in proving this hypothesis.

Figure 3: Average Views Per Hour according to Stream Duration (top 2 values per viewership category are highlighted in green; lowest value per viewership category is highlighted in red)

The only viewership category that benefited from longer broadcasts was the “0–5” category. This is most likely because streamers at these numbers are so difficult to find, so it takes a while for viewers to accumulate there. Also, the longer these channels stream for, the higher their chances are of receiving a host/raid to boost numbers (which is technically true for all channels, but especially so for the smallest ones).

As for the rest of the streamers, it looks like the best durations to stream for (in terms of maximizing views with as little time and effort as possible) are between 0 and 4 hours. Does that mean you would not get new viewers after 4 hours? It does not. You would still get some new viewers (and subscribers and donations as well) after 4 hours, but the rate may slow down after the broadcast drags on for a while. This is similar to how a YouTuber will see a gradual decline in retention as their viewers progress through a video.

The important takeaway is that it is sufficient for a streamer to only spend a couple of hours per day growing their audience. There is no need to do more work than necessary, except in cases of raids, hosts, or when the energy of the broadcast is going fun and strong. If you receive a raid right before ending a 4-hour broadcast, then of course you should go at least another 30 minutes. Use your judgment too in special cases!

Which hours of the day should I stream at?

Now that it has been established how many hours are best, it is time to uncover which hours of the day are best. Is it better to stream mornings, nights, or during the day when streamers and viewers are at peak volume? The data below can answer this with confidence. Be sure to look at the row of graphs/trends at the bottom of Figure 4.

Figure 4: Average Views Per Hour according to Time of Day (top 3 values per viewership category are highlighted in green; lowest 2 values per viewership category are highlighted in red)

For the most part, the trends show some peaks early in the day, followed by some pretty lengthy plateaus for the rest of the day. Technically, this can be looked at in two different ways. On one hand, these trends could reveal that 2:00 AM ET through 11:00 AM ET is just a great period for streaming. On the other hand, they could reveal that the most popular streamers (worldwide) stream during this period. Either way, it means these hours of the day correlate to the highest views per hour. After about 1:00 PM ET, Figure 4 shows only 1 maximum, 2 other highs, and quite a few minimums. Thus, streaming in the morning hours (according to the Eastern Time Zone) is a pretty safe bet to maximize views per hour.

Convert these times to your time zone, and you might just discover the perfect streaming schedule. Also, be sure to use your judgment here as well! If your regular viewers are conditioned to morning broadcasts, you may lose some of them if you switch to nighttime broadcasts. Is this a sacrifice you are willing to make?

Part 3: Finding Gender Bias

Streaming should never be a competition between genders, but gender data is still very useful. Do women earn more views per hour than men do? The results of the 11,000-broadcast study are confident. See for yourself in Figure 5 below.

Figure 5: Average Views Per Hour Sorted by Viewership Range, Outliers Removed (Logarithmic Scale)

Quick Tip: There is a difference between “views” and “viewership.” Read the graph’s axes carefully.

Although trends cannot make conclusions, the data recorded is definitive that women do earn more views per hour at every level of viewership from 0 viewers to more than 5,000 viewers. Due to a couple of shortcomings in the sample, the two sets of points furthest to the right should be taken far more lightly than the rest of the points, but all of the “Female” points were still higher than their respective “Male” points. Why is this? To keep this article as data-forward as possible, only light assumptions will be made based on facts about how Twitch’s demographics actually work. Here is one possibility:

Since there are fewer women on Twitch (35%, compared to 65% men), they would generally get a higher number of views because views would be split between fewer streamers. In other words, it is a statistical truth that “fewer women” means “higher average views” because users seeking female streamers would be divided between fewer female channels.

This is not necessarily something that anyone can control like they would control the game they stream or the time of day they stream at, but an aspiring streamer should still be mildly aware of these statistically-proven gender biases.

Turning Views into Viewers

Part 1: Finding the Right Channel Identity

Realistically, being yourself is the key to success. Unfortunately, this advice will only take streamers so far. There are certain habits and personality traits that the most popular streamers manifest in themselves and their channels. To identify the strongest characteristics of a successful (in terms of growing viewer count) Twitch streamer, a survey was conducted by the author of this article between July 2020 and January 2021. Respondents were asked to rate their channels based on nine (9) unique criterium.

The major findings of this survey can be seen in Figure 6 below.

Figure 6: Results of a Twitch Streamer Self-Rating Survey

The major findings of this survey were that all nine (9) criteria trended positively, but three (3) of them stood out above the others. These were “Brand,” “Consistency,” and “Energy,” as shown in Figure 6. The streamers who a) had the most complete, up-to-date channel pages and biographies, b) streamed on schedule most consistently, and c) applied the highest amount of energy through liveliness, animation, and/or volume tended to average the highest concurrent viewership.

So, for all of the aspiring Twitch streamers out there, get a profile picture and a cover photo, have your broadcasts recorded as Videos on Demand (VoDs) so users can watch them during off-hours, share your interests, link your social media accounts, stream consistently, and give your all every time you hit “Go Live.” The numbers do not lie. This is how proven streamers are converting views into viewers. You can be yourself all you want to be, but that will not entitle you to anything. Fine-tuning, whether a little or a lot, will be necessary to increase your chances of finding success.

Part 2: Finding the Right Moderators

Unlike what many sources would say, moderators cannot just be random viewers whom you trust. The overall behavior tolerance in chat changes from channel to channel, game to game, gender to gender, and ethnicity to ethnicity. Moderators need to be aware of the types of harassment they will face while serving a particular streamer. “How will they know what forms of harassment to expect?” Well, to identify how harassment varies from gender to gender and ethnicity to ethnicity, the author of this article conducted an observational study in 2018 where (a stratified random sample of) 160 channels’ chats were observed for an hour at a time so instances of harassment could be recorded and eventually summed for later analysis.

Note: Not all of the harassment observed was directed at a streamer but rather simply took place in a streamer’s chat. Regardless of the target or intention, harassment still affects the overall sense of community and should not be overlooked or dismissed.

Which types of harassment should I look out for?

The most common forms of harassment that were noticed in various Twitch chats of different-sized channels and games were Homophobia, Racism, Sexism, Gender-Exclusivity, Mental Illness, Religious, Sexual, and General Bullying. Not every form was experienced equally, however. You can see more about this below in Figure 7.

Figure 7: Results Sorted by Type of Harassment Noticed

As you can see, sexual harassment is out of control on Twitch. The blatant sexual harassment should not be an issue for moderators, but the rest of it could be too subtle for an amateur moderator to identify. A lot of the time, sexual harassment is disguised in the form of a joke and is done “with good intentions” between friends (that is just how Twitch users act nowadays). These cases are when a moderator needs to focus on “respect” and “safety.” Sexual harassment is never appropriate, but an experienced moderator who knows the viewers well can usually get a feel of the situation just by observing for a moment. If the intentions are unknown, a moderator should ask the recipient if they are joking around or not, just to be certain whether or not actions need to be taken to restore respect. The messages can always be deleted afterward anyway (without further action), just to enforce a more friendly setting for all types of viewers. Regardless of intentions, though, sexual harassment should be discouraged in all channels.

The rest of the harassment types are never a joking matter (hence why they occurred less frequently during this study), so there is less gray area for moderators to worry about. Unfortunately, Twitch is not an ideal world, and deviants will still target individuals with homophobia, racism, sexism, misogyny, bullying, etc. because they know they can escape with nothing more than a “slap on their wrist” (i.e., a ban or timeout).

Do different genders experience harassment differently?

While moderators should know what to be on the lookout for, they should also know who on the platform is most vulnerable to harassment. Figure 8 and Figure 9 below can offer some assistance with that.

Figure 8: Results Sorted by Streamer’s Gender

Figure 9: Results Sorted by Streamer’s Ethnicity

To summarize, female streamers faced almost twice the amount of sexual harassment and more than three times the amount of gender exclusivity that male streamers faced. Male streamers, on the other hand, faced almost six times the amount of racism and almost twice the amount of homophobia that female streamers faced.

Looking at the data from the perspective of ethnicity, a different story is told. Within the scope of this observational study, white streamers saw almost double the amount of sexual harassment than streamers of color. On the other hand, streamers of color experienced almost double the amount of general bullying and almost three times the amount of homophobia.

Ultimately, all that can be said is every streamer faces different forms of harassment. Some will be predictable, which is just one of the many sad truths of (virtual) society, and others will not be. For the more predictable forms, streamers should work with their moderators to identify certain hateful words and phrases that will inevitably be said in their chats, so those words and phrases can be blacklisted (automatically removed from the chat) before they ever become a problem. Streamers have the tools available to them, and they really should use them. Safety is crucial not only to turn views into viewers but also to keep the entire Twitch ecosystem healthy.

Turning Viewers into Followers

Part 1: Finding Demographic Bias

There is a multitude of ways to get viewers to press the “Follow” button, but measuring their success rate is tricky and requires a lot of testing. Ideally, an experiment would be done where some channels would be given specific titles, vibes, and “Follower Goals,” and other channels would be set up with different ones. Unfortunately, no such experiment has been set up to manipulate certain channels and record their follower rates. What has been set up instead is an 18-month observational study to see if demographics (that are not necessarily “manipulable”) like relationship status, ethnicity, and gender, influence the follower rates of Twitch channels.

Figure 10: Follower Growth Trends per Relationship Status

Figure 11: Follower Growth Trends per Ethnicity

Figure 12: Follower Growth Trends per Gender

Note: To keep these observations as statistical as possible, only minor speculations will be made about each of the graphs above. Keep in mind that these are mere educated assumptions to explain trends. They are not facts. They are possible explanations based on how Twitch actually, certainly, provenly works.

First, relationship status is not often a topic of discussion between streamers and viewers. Many streamers, in fact, (rightfully) prefer to keep their personal life separate from their streaming career. This means a single streamer would be indistinguishable from a streamer who is in a relationship, which explains why the growth rates are so similar.

Second, since the majority (71.5%) of Twitch users are white, it may be easier to find and follow white streamers due to their abundance on the platform. This could explain why white streamers saw a small advantage in follower growth.

Third, since the majority (65%) of Twitch users are male, perhaps they are seeking out female streamers to watch and follow as though Twitch is a dating website (which does happen often enough to be taken into consideration). This could also come down to the fact that there are fewer female streamers on the platform, so follows would be more abundant among them. Whatever the case, the follower growth for women was much more rapid than that of men.

Conclusions

The easiest way to provide a conclusion for all of the incredible information provided by this article is by bulleting every major point. In chronological order of how the data was presented in this article, here are the major trends that the data suggests:

Getting Views:

• Based on numerous criteria, some Twitch categories are better to stream than others. For October 2021, the criteria suggested that League of Legends, New World, Dota 2, Counter-Strike: Global Offensive, Valorant, and Call of Duty: Warzone were among the best categories for an aspiring Twitch broadcaster to stream. This changes from month to month, however.
• Despite the larger influx of streamers on the weekend, the data suggests that it is worth an aspiring streamer’s time to stream on the weekend.
• For streamers with 0 to 5 concurrent viewers, it is beneficial to stream for long periods. For streamers with 6 to >5,000 concurrent viewers, 0 hours to 4 hours seems to be the sweet spot to maximize views per hour (the most views for the amount of time/effort applied).
• In general, the higher numbers of views per hour occur between 2:00 AM Eastern Time and 11:00 AM Eastern Time.
• The data is confident that female streamers earn higher views per hour than male streamers do.

Turning Views into Viewers:

• Survey data suggests that the three most important criteria for aspiring Twitch streamers to meet are Energy, Consistency, and Brand.
• The results of an observational study suggest that sexual harassment is, by far, the most abundant form of harassment on the Twitch platform.
• The data suggests that the chats belonging to “female” streamers experience more sexual harassment and gender exclusivity than the chats belonging to “male streamers,” so a streamer should acquire moderators accordingly.
• The data suggests that the chats belonging to streamers “of color” experience more general bullying and homophobia than the chats belonging to “white” streamers, so a streamer should acquire moderators accordingly.
• As a moderator, it is important to respect everybody, greet every username, keep the discussion going, and remember as many details as possible about viewers.

Turning Viewers into Followers:

• According to an observational study, there is no distinguishable difference between the follower growth of aspiring Twitch streamers who are “single” and those who are “in a relationship.”
• The data suggests that there may be a slight advantage for the follower growth of aspiring Twitch streamers who are “white” as opposed to those who are “of color,” but it is negligible.
• The data suggests that there is a pretty heavy advantage for the follower growth of aspiring Twitch streamers who are “female” as opposed to those who are “male.”

In the end, numerous factors go into deciding who succeeds and who fails in terms of content creation on Twitch. Some of them are within a broadcaster’s control, but many of them are not. If it would not hinder your happiness on the platform to stream popular categories at specific times of day on specific days of the week for a specific number of hours each day, then you might want to at least consider doing what the data suggests. It may not reveal to you a 100%-guaranteed, streamlined path to success, but it might give you that added nudge you need to climb to the next rank. That boost alone is invaluable.

The intended audience for this elaborate series of research studies consists of aspiring Twitch streamers who are aiming to take their content to the next level. If you read up to this point, I assume that is your mission, and I wish you the best of luck! This is no guarantee of success, but data can be very useful for predicting the future of your channel. The rest is on you, however. You know your channel better than anyone else, so use your best judgment alongside the data presented.

Also, all of the data presented is currently being shaped into a “calculator” that a streamer can use to determine their best streaming options for a particular month. All that will be required are a few simple inputs as shown in the following screenshot of an earlier version of the tool. Ideally, it will be ready by this summer, so stay tuned! It is extremely useful.

Message from the Author: The calculator is now available! Thank you for patience! You can try it out here: Twitch Exposure Calculator

If you enjoyed this article and would like to read some of my others, you can find everything here. A lot of it pertains to the likes of Twitch and other social media, but you will find some miscellaneous content in there as well (life, sports, television, etc.).