I mean it's important to note that there is a massive number of people who are just unreported. The people with mild cases (as would tend to be the majority) usually aren't going to confirm they have it or not. So the death rate and everything else is all the reported deaths as a percentage of the reported cases...which is likely well above the real death rate as the reported cases won't include loads of people who just have mild symptoms. Not like everyone with symptoms equal to a mild cold will go test themselves, that's impossible and would cripple the healthcare system even more (it's already happening to some extent where loads who have mild cases are going in out of panic).
But all diseases are underreported. The best any statistician can do is to ensure they're using similar metrics to compare them.
So covid-19 might have a cfr of 2% with a bunch of unreported cases. But then flu will have a cfr of 0.2% and also have a bunch of unreported cases.
The only way to draw any reasonable equivalence is to assume that the data is going to be incomplete in every outbreak.
If we're talking about accurate numbers of infections versus deaths, I agree that the published figures are going to be wrong. But if all we're comparing is one disease to another in order to get an idea of severity, the only thing you can assume is that the data is flawed across the board.
As an indicative number, 2% is a helpful stat to work from.