We always add a little bit extra to the boss values before rechecking the fight and releasing it live.
The team responsible for balancing boss fights does so without debug commands and at the appropriate item level, employing available materia, foods, and medicines while experimenting with mitigating actions and various job compositions. Yet we recognize that player skill far exceeds our own. If we were to ship content with the same values which challenged our battle team, the top raiders would be deprived of that by-the-skin-of-your-teeth victory in the initial week of release.
Based on the team's skill and our experience, that "little bit extra" usually translates to:
Balance test clear values +1-2% HP
The final values account for other mathematical factors too, of course, such as the
estimated damage the party could deal from the moment of victory to the end of the time limit, as well as total burst damage potential based on frequency and amount inflicted. The numbers these various calculations are based on are what we were referencing when we said that our team’s performance was higher than anticipated.