My brother has had 2 major viral videos that have over 3,000,000 views each, and they've made him quite a bit of money. He really doesn't love to get into analytics like I do, but I did get to see his stats and info. I think I have a good idea on what YouTube is looking for at least from a numbers perspective.
Basically, for most videos to go viral, you need a conduit for views thru one of YouTube's main impression gateways.
Those are:
1) YouTube Searches that see insane growth on Google Trends/ect (suggesting massive spiked interest in a topic that you already ranked well on. Think Among Us if you had a video ranking high in May when it grew in August).
2) Browse Features
3) External Virality (Dad How Do I would be a good example - news networks talked his channel up, or that one Mexican grandma that was cooking on a primitive hot stone to show off local cultural fare)
Now, the first two are pretty unpredictable: You have no idea when a game is randomly going to become popular outside of the big ones like, say Cyberpunk 2077 or other AAA releases.
That basically leaves Browse Features as a prominent way to go viral, and that one is 100% controlled by YouTube's algorithm.
Browse Features are controlled by 3 main components (I believe):
1) Clickthru Rate (higher is always better)
2) % Watch Time (higher is better, and there's a kicker for mid-roll ads too)
3) Channel Monetization
It essentially runs off game theory/the social dilemma: YouTube wants to recommend the best content to as many people as possible. If no one will click on a video, then there's no way they are going to feed a video from any one, millions of impressions.
In the case of my brother's viral videos, I can tell you that his Browse Feature CTR was above 15% for multiple days for non-subscribed members. Average Watch Time was over 6 minutes on an 11min video.
YouTube then fed his video roughly 15 million impressions over the course of a week. This resulted in +50,000 subs (or more), and any follow-up content getting 100,000+ views whereas videos prior to this were lucky to get 10,000 in their lifetime.
That was in October last year. He had another video spike in January and it did about 4,000,000 views in 2 weeks. I saw the numbers, and at one point his YouTube Studio app showed 1,100,000 channel views in just 48h - crazy. On that video, the metrics were just about the same. 17% CTR, 7min watch time on a 13min video.
Why did his videos stop getting hits? Simply put, the CTR came crashing back to Earth. The impression windfall stopped when the CTR dropped below 12%. It ran out of people, more or less, to watch it.
I imagine everyone's hard-pressed to figure out how to catch that viral lightning in a bottle, but there's a rhythm and rhyme to it, and its main driver is user engagement. If you can capture it, you're going to do well. Now how you win that battle is for you to figure out.