A hard bounce means stop. The address is invalid, the mailbox is gone, or the domain does not accept mail for that user. Retrying achieves nothing and actively damages your reputation, because repeatedly mailing addresses that do not exist is one of the clearest signals of a poorly maintained list.
The number that matters
Sustained hard bounce rates above roughly 2% signal a list-quality problem. Above 5% you are actively damaging reputation, and providers start paying attention. Amazon SES, for instance, places accounts under review at around 5% and can pause sending entirely at 10%.
The misclassification problem
This is the part that costs people real subscribers. Not everything reported as a hard bounce is one.
Block bounces — where a receiving server rejects your message because of reputation, content or policy rather than because the address is invalid — are frequently lumped in with hard bounces by naive processing. Suppress those and you have permanently removed a valid, engaged subscriber for a problem that was yours, not theirs.
The distinction lives in the SMTP response. A 5xx that says the user is unknown is a hard bounce; a 5xx that says the message was blocked for policy is not. Treat them the same and your list quietly shrinks for the wrong reasons.
Why this is worse across several brands
Bounce processing is usually assessed at the account level by your provider, not per brand. One brand importing a stale list can put the entire account under review, silencing every other brand you run through that connection.
That is the concrete argument for keeping suppression enforcement in the path for every brand rather than trusting each one’s list hygiene independently — and for importing suppressions before contacts on any migration.