The issue is that these sample sizes can be son inaccurate they lead to data that is way far off the mark. For example on Overbuff in GM, Mei has a pickrate 5x that of Genji’s. But the devs stated they actually have the same pickrate in GM. And there’s probably a lot more examples of this, but we don’t know because we don’t have real stats.
The issue is it’s really difficult to give balance suggestions when we don’t have data or stats to even work with to see how the meta is shaping and which heroes are dominating what ranks. It also helps to use the data to support your claim in which you might be calling a hero out on being overpowering or weak.