Brent Ozar has a step-by-step guide explaining what to do when you CHECKDB reports corruption:
This sounds paranoid, but as an example, here’s a corruption case I had recently: shortly after detecting corruption, the team realized they’d have to revert to a backup of the database from a few days ago. Rather than telling users about that possibility, they let the users keep adding data into the already-corrupt database while the DBAs did troubleshooting. Several days later, as the corruption got worse, even Microsoft couldn’t repair the corruption – and the affected tables went completely offline, permanently. If the users would have been alerted earlier, they could have avoided even more data loss.
Good advice. If you have Pluralsight, I recommend Paul Randal’s course on database corruption. Watch that ideally before you have corruption…