Moderator calibration is the process of making sure every moderator applies the same platform policies the same way, so borderline cases don't get decided differently depending on who's reviewing. It usually means moderators review the same sample of content and compare decisions to find and resolve disagreement.

‍

‍Why it matters

Even with detailed written policy, two moderators can reach different conclusions on edge cases the policy doesn't clearly cover, such as sarcasm, cultural context, or content right on a policy boundary. Without calibration, that disagreement stays invisible, and identical content gets removed by one moderator and left up by another. Calibration brings it into the open before it shows up as a pattern in accuracy numbers, and it matters more as the team grows.

‍

‍How a session runs  

A session picks a batch of representative content, including known tricky cases. Each moderator reviews it independently, then the group compares results and works through every mismatch against the policy itself. The resolved interpretation gets written down so it holds going forward, not just agreed verbally in the room.