GAO Schedule Assessment: The 10 Best Practices, Explained
Federal agencies, transit authorities, and large infrastructure owners grade schedules against the GAO Schedule Assessment Guide. It asks harder questions than the DCMA 14-Point, and a schedule that sails through DCMA can still fail it.
The U.S. Government Accountability Office wrote the guide to answer one question: can this program's schedule be trusted to predict its finish date? Reviewers apply it to contractors on federal work during oversight. The guide sorts a reliable schedule into four characteristics and grounds each one in specific best practices. Ten in all.
The four characteristics
GAO calls a schedule reliable when it earns four labels: comprehensive, well-constructed, credible, and controlled. Each label rests on best practices you can test against your own file. Miss enough of them and the reviewer downgrades the characteristic, then the schedule.
Comprehensive: the schedule holds all the work
- Capture every activity. Every piece of scope belongs in the file, including owner actions, permits, long-lead procurement, and level-of-effort work. Reviewers compare your activity list to the WBS and the contract. A gap reads as risk you never planned for.
- Sequence every activity. Give each one a predecessor and a successor so the network drives the dates. An activity with dangling logic floats free and corrupts the forecast around it.
- Assign resources. Load the labor, equipment, and cost the work needs. GAO leans on resource-loaded schedules because they support earned value and a cost forecast a reviewer can check.
- Set durations you can measure. Keep activities short enough to status with confidence. A 90-day activity hides whether the crew is on pace or three weeks down.
Well-constructed: the logic works
- Trace it horizontally and vertically. Horizontal traceability means the logic runs start to finish with no breaks. Vertical means every activity rolls up to the WBS and the milestones above it. A reviewer follows a delay through the network and expects it to land somewhere that makes sense.
- Confirm a valid critical path. One continuous chain of driving activities should connect the start to the contract finish. Open ends, stray constraints, and missing logic break that chain and hide the path that matters.
- Keep total float reasonable. Negative float says the plan cannot meet its dates. Float in the hundreds of days often means a missing relationship. Both draw questions from a reviewer.
Credible: someone tested the dates
GAO wants a quantitative schedule risk analysis behind the committed date. A single deterministic finish assumes every activity runs to plan, and no project does. Run a Monte Carlo simulation across your durations and logic and you get P-level dates. When the owner pushes on the finish, you answer with a confidence level instead of a promise.
Controlled: the schedule stays current
- Update with actual progress. Status the file on a current data date. No forecast dates sit in the past. No actual dates sit in the future.
- Maintain a baseline. Freeze a baseline and measure execution against it. Without one you cannot show whether the work lands on plan or slips, and the Baseline Execution Index has nothing to compare.
Where GAO and DCMA diverge
The two standards overlap on logic, constraints, float, and durations. A clean network passes most of both. GAO then asks for three things the DCMA 14-Point skips: resource-loading, a schedule risk analysis, and a maintained baseline with earned-value support. A pure CPM logic schedule that aces DCMA can stall on those three. Find out which standard the owner uses before you build the schedule, not after they hand back the review.
ARGUSFORGEScore it against both before the owner does
Drop a .xer onto ArgusForge and it grades the schedule against the DCMA 14-Point and the GAO best practices in one pass, with the flagged activities behind every finding. Run the Monte Carlo tab for the risk analysis GAO asks for. Everything runs 100% offline; your schedule never leaves your machine.
See ArgusForgeRun it before the review
Reviewers grade the same file you build, so score it yourself first. Start with the four findings that sink most schedules: a broken critical path, negative float, no baseline, and no risk analysis. Clear those and you walk into the review with a schedule that answers its own questions.
For the checks that overlap most with GAO's logic requirements, read our field guides to leads and negative lags and the full DCMA 14-Point breakdown.