Turns out I have been dealing with trial reels the wrong way.

I thought trial and normal reels have underlying similar mechanisms, so time-wise they are comparable, but it is not that way.

The original normal post has more context, and trial reels start from a clean slate.

So, if I compare the performance of two at both being uploaded 12 hours ago, I am making a big mistake.

I should compare the 12-hour performance of a normal reel with a trial reel that has been up for way longer than that.

It is also dumb to judge any trial reel before 24 hours, and right now I was just deleting them after 12 hours, so ya I need to fix that.

To be on the safer side, I will compare both trial reel and the normal reel after 4-5 days of running them.

I guess till that time, the algorithm might have trained the trial reel sufficiently, so the playing field might be a bit even.

Instead of over-optimizing when to compare, I will just check a bit later.