I find the way people try to use YouTube videos as metrics of character popularity to be woefully inconsiderate of the circumstances in which major trailers debuted and why some videos may have less views than others. Like, a lot of Smash characters debuted in big Directs that already had a ton of people watching and that's how they saw the announcement trailer as opposed to someone like Min Min or Steve who just kind of debuted in their own specific events with some pre-hype. Joker came in The Game Awards totally unannounced whereas Sephiroth was announced as a "new character reveal for Smash" the morning of. People watching those events are in the millions at times, but they won't directly contribute to watching a newcomer video. Similarly, YouTube's algorithm makes it so certain videos that are not the official video appear at the top of search queues. Like Banjo & Kazooie's trailer appears via GameSpot trailers before the official Smash Bros one if you just search "Banjo Smash". You'd have to theoretically cross reference all the views of these videos as opposed to cherry picking the official trailer to get a somewhat complete understanding of YouTube views on trailers... and even then, it still doesn't consider the live events/streams/etc.
Not to mention, YouTube views can just be the same person watching the same trailer over and over again. Numerous people have said "Oh I watched the Banjo & Kazooie trailer 10 times and I can't stop watching it" or some version of that for a character they like. Sure those are die hard fans, but it's an area that calls into question the validity of the "data" when drawing conclusions.
Like I get people are looking for justification for their arguments and there isn't much in the way of data for DLC fighters, so bad data is something to latch on to, but like I said, I think using YouTube views to draw certain conclusions is entirely misplaced.