Personally, I am always more suspicious of 20.1% in this context when someone is obviously trying to bullshit with fake precision in a context that is inherentl

Personally, I am always more suspicious of 20.1% in this context when someone is obviously trying to bullshit with fake precision in a context that is inherently not that exact. 20% vs 21% is basically meaningless in this context. We can't run it a 1000X or see the counterfactual to know what figure would be more correct. I just take 20% to mean "a good chunk" and much preferable to a style of fake precision you seem to think is better here.