Errors in production
Trust the new right or keep the old wrong?

Right or trusted. It happened to everyone who worked hard enough to create a new version of an existing report. You cannot match numbers, so you investigate and find the new numbers are more accurate than the old ones. At that point, the real problem begins: how do you communicate the news? How do you make users trust the right numbers after years when they confidently used the wrong ones?
The joke comes from new calendar-based time-intelligence features that control how you consider days in previous-period comparisons. There is no easy way to communicate that the new version is more correct than the old one. With AI accelerating migration to new systems, this will happen more often than ever. Brace yourself.