I think you need to understand how tiers work to begin with. By default, I believe most proper legendaries are stuck in uber by nature of base stat total and their unique abilities and moveset.
For a pokemon to be considered viable to move "down" a tier, it needs to have one of the lowest usage rates for Pokemon in that tier. So, say Zamazenta is in Uber. Zamazenta is also the least used pokemon in uber tier in the Ubers metagame. There would then be a behind-the-scenes suspect testing that would experiment to see if Zamazenta could instead be brought down to OU to see if it wouldn't break the metagame there and become the #1 most used pokemon in OU. Sometimes it's not apparent at first and it looks like it works out OK. If that happens, it gets brought into a "BL" borderline tier. It doesn't have what it takes to be a popular pick in the tier above it, but it's way too good for the metagame/tier below it. Assuming it doesn't warp the metagame, it can do a proper tier shift and just be "OU" if it just didn't have what it takes to compete in Ubers. That's how stuff like Celebi, Jirachi, etc moved down tiers when they were originally Uber in the past. They weren't good enough for Ubers with their balanced stats.
The reverse is true for Pokemon too good for their tier. When Garchomp eventually saw usage in half of every OU team in DPP, they had to do suspect testing to ban it into Ubers. Garchomp could compete with Legendaries just fine without warping Ubers, and the lack of Garchomp in OU also helped vary teambuilding, because it meant people also didn't have to take an Ice Shards user on every team (Weavile, Donphan, Mamoswine, etc...). However, that just also left enough of a power gap to ALSO let Salamence shine in Garchomp-less DPP OU, which ALSO inevitably led to Salamence going Ubers as well.>tl;dr:
Tiers (OU, etc) are intended to be (in isolation) a balanced metagame, each, and usage statistics are taken into account when doing ladder testing.