The worst? Data Loss, or corruption, which is worse than "not working".

Likewise, more likely it could just not work well, be buggy and require several updates to get it right, burning out your team's (or your company's) image with the customer. Before you deployed your fix, it's whoever changed the protocol at fault. After that, it's you that can't fix it.

Again, not justifing it taking ages, just saying there's never an excuse to skip QA.