# Understanding Twitch Viewer Retention Statistics

Today we’re going to talk about understanding Twitch Viewer Retention Statistics. We will answer the question, “I know how many followers thought I was cool that one time, but how many should I expect will stay for the long haul?”

To collect this data I sourced the spreadsheet results from my favorite Twitch API reader site, Sullygnome! I snagged the top 500 most watched English-speaking streams over the past week (8/16/19) for the first data set, and then I ran a sample of 400 English-speaking partnered streamers with concurrent viewer averages for the last 7 days of 100-127. That composed the second set of data, hoping it would be more relevant to growing streamers.

I did leave out non-partnered streamers from the data for the second set, just because some of it was a huge mess. Simply put, there’s definitely still botting occurring on Twitch, and I was running into results where people would stream for 2 hours, had 50,000 hours watched, and had 22 whole followers. Obviously suspicious. So, I figured the best way to obtain reliable information was to sort by streams that Twitch itself had already reviewed (partners).

-I calculated retention rate as Concurrent viewership divided by total follower count. Retention is expressed as a percentage of total followers. It simply says, out of X number that followed, Y% are now concurrent viewers. Simple math, reliable numbers.

YOU’RE BORING ME SKULL! FINE. SORRY. F**K! Let’s get into some of the results.

Among the top 500 most watched streams, average retention, or percentage of followers who become concurrent viewers, landed in at a whopping 2.03%. That translates to every 98 out of 100 people who followed those streams, not habitually returning. This is almost exactly as I expected. Retention/engagement on most social media hovers between 1% and 3%. (source) That’s true of Twitter, Instagram, Facebook, etc.

Twitch is no exception to the rule, and remember these are the 500 MOST WATCHED streams across the service. These are the ones attracting attention.

Wanting to more fully understand the reasoning behind why some streams beat the average of 2.03%, I delved deeper into any stream that had an average retention higher than 3%. Here were some of the results:

If you take a quick look, a few things should start to pop here. Tournaments, Embeds, Professionals, and Company-Sponsored streams appear to have the highest retention. Let’s address each:

Tournaments: Tournaments often have a clause in their signups that entrants that they must either stream gameplay, are promoted on each player’s social media, or the promotion is handled by the players’ marketing teams. The promotion works for both the tournament, the players, and the team’s benefit. Tournaments are also typically embedded across their sponsors’ websites, increasing the viewer count further.

Embeds: Some of the streamers in the list are openly sponsored by websites. Inspecting the website typically reveals that the streamer has their stream playing on that website’s page. Each person visiting the website counts as a viewer for that streamer.

Professionals: Some of the streamers in the list are professional gamers, and have signed a contract with a team. In this data set, most of the professional examples were Fortnite, League of Legends, DotA, and Teamfight Tactics players. Professionals not only see the benefit of having above average talent, but have the marketing of their team behind them. These players also tend to be embedded on their respective team websites, and gain widespread popularity/notoriety from tournament participation. (See tournaments above.)

Company-Sponsored: Exactly as it sounds. Bungie’s at the very bottom of this segment of data. Bungie has years worth of marketing behind its Halo and Destiny titles, in addition to managing their own community outreach programs, social media, articles written about their products, embeds, etc. Some other examples are Riot Games (League of Legends), DotA, Call of Duty, and more.

Most of these outliers in the data are TEAM and COMMUNITY efforts. They do not represent single-person amalgamations, or normal-case scenarios. Most noteworthy though, each stream here has a significant marketing effort behind making their name known.

Another category that begins to pop further down is YouTubers who also stream on Twitch. In several cases, YouTubers that maintain large followings are able to transfer some of their viewership over to their Twitch streams. Analysis is still unclear whether the percentage is significant, but it’s common enough in the results to suggest it’s a factor.

So that covers the top 500 most watched. What about streamers closer to our level? Well I ran 400 samples of partners between 100-127 concurrent viewers. The average retention among this subset of data was 0.92% or about 1%.

Conclusion

So there you have it. Typical Twitch retention across the board is between 1-2% for the vast majority of streams, independent of content type, channel age, channel size, main game, demographics etc. If you’ve got retention higher than 2%, you’re bucking major trends. And now you know what to expect when it comes to your own channel!

Anecdotally, most partners have been getting partnered with around 5,000 total followers, and of course average viewership of 75 concurrent viewers or higher. This is well within our range for our results here, lending more credibility to our research.

Sadly, the limitation on this data is the unavailability of unique view counts. There’s not easily obtainable information on the number of people that come to a channel and never follow it, so it’s very difficult to tell what behaviors or features cause more following. A question for another time.

Future practical application of this study: We know how many follows we’d need for a certain concurrent level. If there were a tool to estimate our follower gain per hour, we could in theory reason back to how many follows we would require to reach specific goals, and by analyzing the rate of follower gain, you could plan out whether your goals could map out the time required. Again, a pursuit for another time.