I am a big fan of MatPat, the guy behind The Game Theorists and (my favourite of his channels) Film Theorist. But as well as writing video essays about games and films he also has made lots of videos about how YouTube works. With his experience of building 5m and 8m subscriber-strong channels I figure it’s worth looking at what he has to say.
There are at least 8 videos on how YouTube works in his collection, and they’re worth watching in full, but here are the key points as I see them. Some of it’s quite basic, but other stuff (like his argument that publishing too frequently can backfire) is interestingly counter-intuitive.
1. Subscribe ≠ subscribe
MatPat wants popular YouTubers to stop complaining that their subscribers aren’t seeing their videos. Sure, when you subscribe to a magazine you expect it to be delivered through your door every week without fail. But “subscribe” on YouTube is just an indicator to YouTube – one of many – that tells it you’re interested in that channel’s output.
2. “Subscriber burn”
The ideal channel, as far as YouTube is concerned, is one that viewers engage with 100% – meaning they watch everything produced. If you stop engaging with videos, the score for that channel (as far as that user is concerned) will go down. If it gets too low, YouTube will stop suggesting your videos to that user; they will have been “burnt out” by the YouTube algorithm. This is how a YouTuber like PewdiePie can have over 50m subscribers yet only see hundreds of thousands of views on particular videos.
3. YouTube is a dating service
I like this analogy. YouTube is trying to pair you with videos you want to see. If it fails to do that you won’t use the service.
The problem is there’s more videos to see than you could ever watch even if you sat there 24/7. For that reason YouTube sifts through your subscribed channels and other videos on YouTube and picks out ones it thinks you most want to watch. YouTube decides these recommendations (on the homepage and next to videos you’re already watching) based on two “neural networks”:
The first is the candidate generator which decides the videos to recommend. Then comes the ranking filter which decides the order of those recommended videos.
Factors for the candidate generator include:
- What you’ve searched for recently
- Where you’ve spent the most watchtime
- How recent the video was uploaded (its “freshness”)
Factors for the ranking filter include:
- How long have other people watched it?
- Are you subscribed?
- How many of that channel’s previous videos have you watched?
- How good the thumbnail is (apparently YouTube has had a thumbnail algorithm for years)
A bunch of things follow from all this:
4. Remember YouTube is a search engine
YouTube is the world’s second most popular search engine and search is how many people end up watching a video. He argues that part of PewDiePie’s high-profile troubles in 2016 were because he wasn’t titling his videos with searched-for terms or topics. Instead of making content about video games which are always being searched for he was making videos with titles like “SOMETHING HAPPENED”. When he did do something that aligned with what people are searching for – a Halloween makeup tutorial – it did really well.
For that reason – if you weren’t already – make sure your titles, description and metadata talk about the subject your video is about.
5. Go for long videos
MattPack argues that “watch-time” is incredibly important. Create things that viewers are going to want to watch for a long amount of time.
Apparently, the change came in March, 2012 when YouTube switched from prioritising clicks to prioritising watchtime, which led to the domination of Let’s Play, and gaming channels, and was behind the success of Minecraft.
Obviously this can backfire if your videos become boring and viewers start skipping them, but it’s worth remembering that YouTube sees itself competing with other video providers like Netflix and Amazon, and wants to encourage videos that encourage immersive viewing behaviour.
6. Back catalogue views have seen a decline
Starting in the summer of 2016 MatPat argues that YouTube started to emphasise daily engaged users as part of its effort to be a daily habit. This meant that it started to emphasise the newness of videos and that has meant a decline in the performance of old videos, often a key way channels make money.
7. Consistent format + new trend = good content
Because of the emphasis on ‘freshness’ a good YouTube strategy is to have a format that allows you to comment on popular things quickly. For example, all the “try it” formats are ideal for jumping on a trending topic, whether it’s fidgit spinners or slime.
8. You can ignore likes and comments
According to MatPat the amount of likes and comments a video gets are “largely meaningless”, and has little impact on how likely YouTube is to recommend it to viewers.
9. Know who your target audience is
“When channels we consult for say their audience is everyone […] that is a big red flag to us […] by trying to appeal to everyone you’re a lot more likely to reach no-one.”
10. Too many uploads can backfire
MatPat argues that uploading too frequently can be bad because if there’s too much content, users will start skipping videos and YouTube will take this as a sign that you don’t want to see that content as regularly. He says he’s recommended some channels cut their uploads in half, and as a result have seen their views double.
11. …But upload frequently
At the same time he says that since July 2016 when YouTube tweaked the algorithm to favour daily engaged users, it’s important to upload frequently. They did this because they wanted to make watching YouTube part of people’s daily routine, similar to the way that people switch on Netflix or the TV at the end of the day.
He argues that channels like Epic Rap Battles of History (which use a season approach similar to television) walk a very dangerous line. If the first episode of the season doesn’t work, all the subsequent episodes will struggle.
12. Consistency is key
MatPat argues another reason PewDie Pie suffered was his content changed a lot. It went from video game playthroughs to personality-led vlog pieces. He says this was also what hit the cover artist scene really hard in 2015. “One channel started vlogging between songs. Others bandwagoned to follow suit and as a result all the channels died. Because the vlogs started to water down the content that everyone was coming for: the cover songs.”
13. Use your popular uploads to support your new shows
MatPat’s channel publishes a range of different shows, the most popular of which is Game Theory. However, because it takes a long time to produce (100 hours he says) he has to have other content in between.
However these new shows aren’t as popular as Game Theory. If he does too many of these shows his subscribers will be burnt out and stop being recommended his content.
For this reason he’s very careful about how he does things. To support a new show – e.g. Smash History – he’ll release an episode just after a more popular Game Theory episode. That way it’s more likely to be seen, and less likely to cause subscriber burn.
14. Being international is an advantage
In one video MatPat argues PewdiePie’s domination was helped by the fact that the YouTube algorithim promotes videos created locally. So PewDie was promoted in Sweden where there was less competition, and later in Italy, after he moved there.
15. We hate clickbait but we fall for it anyway
Your brain is wired to dislike unfinished stories so when someone writes “I can’t believe this happened” or “Why I’m deleting my channel”, you get a shot of dopamine by clicking on it – in fact he argues you get the dopamine even before you know the end of the story. That’s why even though clickbait is annoying, people don’t actually give clickbaity videos the thumbs down. We like it.
What’s more, because it’s a thing of chance (sometimes the video is disappointing, sometimes not) it’s extra addictive. It’s the same reason all games of chance are fun. Sometimes we’re disappointed but when we’re not, the dopamine hit is all the higher.
16. Creative burnout means media companies own the future
With such emphasis on daily engaged users, it will be really hard for individual creators to keep making daily videos without suffering creative burnout. Going on holiday for two weeks will hurt your views and income, so only media companies like news organisations that are used to daily output will survive.
17. Are you ready for the next algorithm change?
In 2012 YouTube switched from prioritising clicks to prioritising watchtime. This meant some YouTubers going from 20m views a month to 3-5. Are you ready for when that happens again? What if YouTube switched from an algorithm that looked at watchtime to one that looked at most revenue from advertising? Is your channel ready for that?
Here’s the full list of videos:
- Game Theory: How PewDiePie Conquered YouTube July 2013
- Game Theory: Yes, PewDiePie. YouTube IS Broken Dec 2014
- Game Theory: Will 2015 be THE END of YouTube Rewind? – Dec 2015
- Game Theory: Is YouTube Killing Pewdiepie and H3H3…and Everyone? Dec 2016
- Game Theory: The REAL Reason YouTube is Broken Dec 2016
- Game Theory: How Minecraft BROKE YouTube! May 2017
- Game Theory: Beyond Fidget Spinners – How to Create a YouTube Trend June 2017
- Game Theory: The Viner Invasion of Jake Paul and Logan Paul! 25 June 2017