Every construction schedule shows a single completion date. That date looks precise, but it isn't. It's the result of hundreds of duration estimates, each of which could easily come in shorter or longer. A schedule risk analysis asks a more honest question: given everything that could go wrong or right, how likely are we to actually hit that date? For owners making commitments to tenants, lenders or a board, the answer is often uncomfortable, and that's exactly why it's useful.
This guide explains schedule risk analysis in plain terms, how a Monte Carlo simulation works, why schedules are almost always optimistic, and how owners can use the results without needing to run the software themselves.
Why a single date is misleading
A CPM schedule treats every duration as a fixed number. Excavation takes 15 days. Structure takes 60. Real work doesn't behave that way. Excavation might take 12 days if the ground is good and 22 if it isn't. Structure depends on weather, crane availability, inspections and crew productivity.
When you add up dozens of uncertain durations along several paths, the uncertainty doesn't cancel out. It tends to push the finish later. That's partly because delays usually have more room to grow than savings do. A task scheduled for ten days might finish in eight, but it could also take twenty. And it's partly because of something called merge bias.
Merge bias: why parallel paths make schedules late
Wherever several paths of work converge on one milestone, the milestone can't happen until the latest path finishes. The U.S. Government Accountability Office's Schedule Assessment Guide (GAO-16-89G) uses a simple example to show what that does to the odds.
Imagine a commissioning milestone that needs two start-up test paths finished: electrical and plumbing. Each path has a 60% chance of finishing on time. The milestone is only on time if both are, so its chance is 60% × 60%, or 36%. Add a third path, such as heating and air conditioning, also at 60%, and the milestone's chance of being on time drops to about 22%.
View the data
| Converging paths | Chance milestone is on time |
|---|---|
| One path | 60% |
| Two paths | 36% |
| Three paths | 22% |
Building projects are full of these merge points: building enclosure, permanent power, elevator turnover, occupancy inspections. GAO notes that the chance of meeting a merge point falls as more paths converge, and that merge bias depends on the risk on each path, the free float before the milestone, and the number of merging paths. A deterministic CPM schedule can't show any of this. A risk analysis can.
How a schedule risk analysis works
GAO describes schedule risk analysis as one of its ten best practices for project schedules. It says an analysis "starts with a good critical path method schedule." The basic steps are:
- Start with a sound schedule. If the logic is broken, the analysis will be too. Open ends, hard constraints and missing owner activities have to be fixed first.
- Capture duration uncertainty. For key activities, estimate a range instead of a single number, typically an optimistic, most likely and pessimistic duration. This is called a three-point estimate.
- Add specific risks. Link risks from the risk register, such as poor soil, late equipment, a slow permit or a winter pour, to the activities they would affect, with a probability and an impact range.
- Run a Monte Carlo simulation. Software recalculates the schedule thousands of times, each time picking random durations from the ranges and randomly deciding whether each risk occurs. Each run produces a possible finish date.
- Read the results. The thousands of finish dates form a probability distribution. From it, you can read the chance of finishing by any given date and see which risks and activities drive the most delay.
GAO's guide says an analysis should be performed before a baseline is set and repeated periodically as the schedule is updated.
Reading the results: P50, P80 and the S-curve
The main output is an S-shaped cumulative probability curve. The horizontal axis is the completion date. The vertical axis is the probability of finishing on or before that date.
- P50 is the date with a 50% chance of being met: a coin toss.
- P80 is the date with an 80% chance of being met. Many organizations use P80, or something close to it, for commitments they can't afford to miss.
- The deterministic date, the single date the CPM schedule shows, often lands well down the curve, with a low probability of being met.
View the data
| Point | Completion | Probability of finishing by then |
|---|---|---|
| Schedule date | Week 52 | 6% |
| P50 | Week 56 | 50% |
| P80 | Week 58.2 | 80% |
The gap between the deterministic date and the P80 date is a measure of how much schedule contingency the project needs. GAO's guide describes using the simulation to determine the contingency, or reserve of time, needed for a chosen level of confidence. GAO's own analysis of one federal construction program reported a range between the 50th and 80th percentiles rather than a single date, which is the right way to present the result.
What the analysis also tells you: risk drivers
A good risk analysis doesn't just produce a date. It ranks the risks and activities that contribute most to the spread. That ranking is often more useful than the curve itself.
If three risks account for most of the gap between P50 and P80, that's where mitigation money should go. In one case study in GAO's guide, risks were prioritized by how many calendar days mitigating each one would save. That's a language boards understand.
Where risk analyses go wrong
A schedule risk analysis is only as good as its inputs. The common failures:
- A weak base schedule. Simulating a schedule with broken logic produces confident-looking nonsense.
- Ranges that are too narrow. Teams anchored on the existing durations give optimistic and pessimistic values that are barely different. GAO's guide stresses collecting unbiased risk data, including anonymously where needed, for this reason.
- Ignoring correlation. If cold weather slows concrete, it slows all the concrete. Treating each pour as independent understates the risk.
- Changing durations to hit a preferred date. GAO specifically warns against adjusting durations simply because an earlier finish is preferred.
- Running it once. An analysis done at the baseline and never repeated tells you nothing about the risks you're facing now.
Cost risk follows schedule risk
The same simulation approach can be applied to a cost estimate, with ranges on quantities, unit rates and escalation, plus specific cost risks. On construction projects, though, the two can't really be separated. A late project costs more: extended general conditions, extra supervision, escalation on work that hasn't been bought yet, and sometimes lost revenue for the owner.
An integrated cost and schedule risk analysis models that link directly. Each simulated finish date carries its own time-related costs, so the result is a range of total costs that reflects schedule risk as well as pricing risk. For owners setting a contingency, this is far more defensible than a flat percentage. It also explains to a board why a three-month delay risk is a budget issue, not just a scheduling one. Our guide to construction contingency covers how to turn that range into a contingency amount and a drawdown plan.
When owners should ask for one
Not every project needs a full Monte Carlo analysis. A small fit-out with a simple sequence and a flexible opening date may not justify the cost. It's worth the effort when:
- The completion date is tied to a lease, a store opening, a school year or financing terms, where missing it has a real cost.
- The project has many parallel paths converging on a few critical milestones.
- There are significant known risks: complex ground, long-lead equipment, phased work in an occupied building, or heavy dependence on permits and utilities.
- A board or lender is asking how confident you are in the date.
On programs we've supported, the most valuable analyses weren't the most sophisticated. They were the ones done early, on a well-built schedule, with honest input from the people doing the work.
What owners should do
- Stop committing to the deterministic date. When you promise a date to tenants or the board, ask what probability it has. If nobody knows, that's the first problem.
- Ask for a risk analysis before baseline acceptance on projects where the date matters. It can be done by the contractor, the owner's scheduler or an independent consultant.
- Present a range, not a date. Report a P50 and a P80 to your board, and explain what drives the gap.
- Fund mitigation for the top risk drivers. Spend where the analysis shows the most days saved.
- Repeat the analysis at key points, such as before major procurement, after enclosure, or when the critical path shifts.
The bottom line
A single completion date is a forecast pretending to be a fact. Schedule risk analysis replaces it with something more useful: a range, a confidence level and a list of the risks that matter most. Owners who make commitments on that basis are rarely surprised, and they can explain their position when the board asks.
Want an independent view of how confident you should be in your project's completion date? Book a 30-minute call with Benjamin R. Consulting.
Sources: U.S. GAO, Schedule Assessment Guide: Merge Bias and Schedule Underestimation; U.S. GAO, Schedule Assessment Guide: Conducting a Schedule Risk Analysis; U.S. GAO, Best Practice 8: Conducting a Schedule Risk Analysis; U.S. GAO, Schedule Assessment Guide: Best Practices for Project Schedules (GAO-16-89G). The S-curve example is illustrative. This article is general information, not legal advice.