>>926197
Since we're numbers, better use numbers, right? :P.
>TL;DR
Twitch is heavily desktop still (80%+ compared to YT's 50%-). If we use pre-lockdown numbers to "project" where Twitch's viewership hypothetically should be at, it's not too far off currently, so the site actually dying is not happening anytime soon. So, in the current situation, it's still a decent idea to try and take advantage of the site to expand and grow viewership. Rich people also like owning large media. If they're willing to buy up newspapers and cable news networks, Bezos is not going to give on Twitch anytime soon.
>Sullygnome - establishing that yes, Twitch is mostly gaming-focused
No, you're right, Twitch still should be best for gaming and gaming-related streams.
For the sake of simplicity, Just Chatting, IRL, and Special Events, were factored out of total hours watched.
Total Hours Watched 1,360,804,753 from
https://sullygnome.com/
3 Categories I named 249,890,390 from
https://sullygnome.com/games/watched
Those three categories make up about 18.36% of total hours watched. There's of course, sports, ASMR, Art, music categories, but the point here is to establish that Twitch is still a majority gaming-focused site. Plus, both special events and just chatting are usual heavily gaming-related (podcasts, tournaments, directs, etc.).
>TwitchTracker
Next, I grabbed monthly viewership numbers from TwitchTracker. I created two graphs with those average viewership numbers. Green dots represent data from about before lockdowns (anything before March 2020). Lockdowns started in the US on March 15, but for the sake, March as a whole is treated as "After Covid". Graph 1 has a simple linear projection using "Before Covid" numbers. Graph 2 has a polynomial projection. Included both because they had the best R-squared for the memes.
Ignoring the projection lines, we can just see from the data points that lockdowns created an artificial bump that peaked mid-2021. Arguably, based on the linear graph, Twitch would have "approached" where it should have been had lockdowns not happened ignoring the results from September and October. If the site is "declining", it's only falling below "expected" growth the past 2 months. Everything else just represents Twitch not properly retaining its lockdown gains.
>Desktop Market
This site's data analysis ends early 2023, but it does provide its links. Click on the semrush links gets the following. Data looks to end September 2025. The website traffic by country is probably for the month?
>Twitch - https://www.semrush.com/website/twitch.tv/overview/
US is 87.25% desktop from 246.66MM devices, the other top 4 are between 86-92.5%
>YouTube - https://www.semrush.com/website/youtube.com/overview/?utm_campaign=twitch-users&utm_source=backlinko.com&utm_medium=referral
Of the top 5, US is the highest at 47.95% desktop users of 5.92B devices. The others are anywhere between 6.65% (South Korea at 6.69B) to 38.29% desktop. This means that more than half of users are mobile.
It's not a proper 1-to-1 comparison though, since it compares all users of YouTube to Twitch, not just live viewers.
>Emotes
This represents an interesting opportunity for Cover and holomems. Good emotes should both encourage people to subscribe to the channel and to use them in other peoples' chats. It both trains the viewers to use Cover-affiliated emotes in holochats and freely advertises holos on behalf of Cover. There should be some effort in "substituting" popular emotes just as a means of "controlling" and "nurturing" chat culture. Given the number of holomems, there should be some opportunity to build an insular, self-reinforcing ecosystem.