Every Monday starts with a status update where every project says "on track," until two weeks before the deadline when it suddenly is not. The signs were there: velocity was declining, the same person was a bottleneck on three projects, and a dependency from Team A was quietly slipping. But status reports do not capture trajectory. They capture optimism.

Program managers try to catch this manually: checking each project board, cross-referencing timelines, asking PMs pointed questions. But with ten or more projects running simultaneously, nobody can hold the full dependency graph in their head. The risks that kill projects are not the ones people report. They are the ones that hide in the gap between what people say and what the data shows.