Schedule Risk
Runs CPM forward/backward pass over the schedule, then samples activity durations from a triangular distribution driven by real risk data (late procurement, stale RFIs, pending COs, overdue submittals). Returns p50/p90 completion bands with named drivers and ranked pull-forward actions.
What it does
Schedule Risk is the simulation engine that answers "are we still going to hit the date?" with real math instead of gut feel. It runs a CPM forward and backward pass over the project's schedule_items (with predecessors) to find the critical path, then runs a 1000-trial Monte Carlo simulation. Each trial samples activity durations from a triangular distribution — the engine sets the min/mode/max from the project's live tracker state, not from generic optimism factors.
Drivers that widen the triangle: late procurement deliveries on critical items, stale RFIs without responses past the deadline, pending COs that touch the path, overdue submittals on long-lead products. The output is the p50 (median completion date), p90 (90th percentile completion), and the top 5 named drivers sorted by their contribution to the spread.
The engine also ranks pull-forward actions — specific moves that, if executed, recover the most days of slip. "Expedite RFI 042 response" or "Pull submittal 0903 to in-house review" with the modeled day savings attached.
When to use it
- Monthly owner report — owner wants a defensible p90 instead of "we think we're okay."
- Procurement just slipped a delivery and you need to know if the path absorbed it or got hit.
- RFI aging is creeping up and you need to know which ones actually matter.
- Considering acceleration — quantify the schedule recovery worth chasing.
- Pre-construction phase — baseline the schedule risk profile before contract execution.
When the simulation detects significant slip, chain into Change Order Review with the time-impact context pre-loaded — the engine helps determine whether the delay justifies a time-and-cost change order.
What to upload
Nothing — this is a server-side aggregation engine. It reads schedule_items (with predecessors), procurement_items, rfis, change_orders, and submittal_revisions for the active project. The risk drivers are derived from those tables; there's no document upload step.
What the engine needs is a real schedule. Schedule items must have predecessors set (the CPM needs the dependency graph). Tracker data must be current (stale RFIs and overdue submittals are the drivers). Optional: explicit risk overrides per activity if the PM knows something the trackers don't.
Step by step
Open the engine
Sidebar → Engines → Schedule Risk.
Confirm schedule is current
Banner shows the date of the most recent schedule update. Stale schedule means stale results — refresh first if needed.
Run simulation
15–30 seconds. 1000 trials of the CPM with sampled durations.
Read p50 / p90 bands
Top of the result: median completion date (p50), 90th percentile completion date (p90). Spread between them is the project's risk profile.
Identify the critical path
Engine highlights the critical-path activities. Activities with high near-critical probability also flagged — these become critical if any current critical activity recovers.
Read the named drivers
Top 5 risk drivers with their contribution to the spread. "RFI 042 stale 8 days → 6 days median schedule impact." Each driver links to its source record.
Walk the pull-forward actions
Ranked list of specific moves with modeled day savings. Apply or note the ones you commit to.
Chain forward
Time-impact COs → CO Review. Long-lead risk → Procurement Tracker. LD risk → Pay App Review.
Understanding the results
Median (p50) and 90th-percentile (p90) completion dates with the contract date for reference. Spread between them is the simulated risk profile.
Activities currently on the critical path, with their slack/float and their probability of remaining critical across the 1000 trials.
Top 5 drivers sorted by their contribution to the spread. Each names the specific RFI / CO / submittal / procurement item with its tracker reference.
Ranked list of specific actions with the modeled day savings. "Expedite delivery on PO 0412 → recover 4.2 days median."
Distribution chart of the 1000 simulated completion dates. Shows the shape of the risk — tight cluster vs long tail.
Claude-written 2–3 paragraph summary of the result for the owner report or internal exec brief.
Every control, explained
Run simulationConsumes one run. 15–30 seconds. 1000 Monte Carlo trials.
Risk override per activityManually adjust the triangle parameters (min/mode/max) for an activity if you know something the trackers don't.
Re-run after overridesRe-runs the simulation with the new parameters. Does not consume a fresh run within a 5-minute window.
Chain to CO ReviewOpens CO Review with the time-impact slip pre-loaded for change-order analysis.
Chain to Procurement TrackerOpens Procurement Tracker filtered to the at-risk long-lead items the simulation surfaced.
Chain to Pay App ReviewSurfaces LD risk for the next pay application review.
Export PDFExecutive risk brief with bands, drivers, and recommended actions. Standard format for owner meetings.