I would think differences should only be counted as variants if they were intentional. Well, also unintentional differences if they are objectively quantifiable (such as missing words etc.)

The depth and pressure applied when making a hotstamp is not an intentional design difference, and also isn't a black-or-white difference. You can't objectively say "this slightly-weaker hotstamp is definitely variant I, while this slightly-stronger one is definitely variant II" for example.