Home / Aerospace and defense program execution
Most defense programs already have a Microsoft Project schedule. The problem is rarely the software. It is a file that grew by accident: tasks in manual mode, dates pinned by dragging bars, links that no longer reflect the work. Set Boundaries builds and maintains Microsoft Project schedules that pass the DCMA 14-point assessment, support earned value and give the program manager dates worth defending.
What makes a Microsoft Project schedule defensible?
A defensible schedule is a working model of the program. Every task is linked to what drives it and what it drives, durations reflect the work, constraints are rare, and the critical path the file shows is the one the team recognizes. When those conditions hold, the schedule predicts the finish date and shows the consequence of every slip. When they do not, the dates it produces are opinions that happen to be printed on a Gantt chart.
The Government Accountability Office sets out the widely used standard in its Schedule Assessment Guide, GAO-16-89G, as ten best practices. They apply to any tool, and they translate directly into how a Microsoft Project file should be built.
The ten GAO scheduling best practices, in Microsoft Project terms
| # | GAO best practice | What it means in Microsoft Project |
|---|---|---|
| 1 | Capture all activities | Every piece of work in the statement of work and work breakdown structure appears as a task, including supplier, customer and test events. |
| 2 | Sequence all activities | Every task has a predecessor and a successor, linked task to task, never to summary rows. |
| 3 | Assign resources to all activities | Labor and material are loaded so the schedule can support earned value and show capacity problems. |
| 4 | Establish the duration of all activities | Durations come from the people doing the work and stay short enough to status honestly. |
| 5 | Verify horizontal and vertical traceability | Detailed tasks roll up to milestones and to the integrated master plan, and logic flows across teams. |
| 6 | Confirm the critical path is valid | The longest path runs from the status date to the end through real logic, not through constraints. |
| 7 | Ensure reasonable total float | Float values are believable; large float usually means missing successors. |
| 8 | Conduct a schedule risk analysis | Duration uncertainty is modeled to show how confident the program can be in its dates. |
| 9 | Update the schedule with actual progress and logic | Actual dates and remaining durations are recorded against a fixed status date every period. |
| 10 | Maintain a baseline schedule | A controlled baseline is kept and compared against the forecast to measure performance. |
Which Microsoft Project settings matter most on a defense program?
A handful of options decide whether the file behaves like a network or like a drawing. We set and document them at the start of every engagement.
- New tasks are auto scheduled. Manually scheduled tasks ignore logic entirely, which defeats the purpose of the schedule.
- A status date is set and used. Progress, earned value and invalid-date checks all depend on it.
- Constraint behavior is deliberate. Where a real contractual date exists, a deadline usually serves better than a hard constraint, because it flags the slip without overriding the logic.
- Earned value options match the program. Microsoft Project can calculate earned value from percent complete or physical percent complete, against a chosen baseline. The choice should match how the earned value system measures progress.
- Calendars are real. Holidays, shutdowns and shift patterns are in the project and resource calendars, not handled with padded durations.
- Custom fields carry the structure. Work breakdown structure, control account, work package, earned value technique and responsible manager are stored on every task so the schedule can be filtered, audited and reported.
What are the most common Microsoft Project mistakes?
| Mistake | What you see | The fix |
|---|---|---|
| Dragging task bars | Dates stop moving when predecessors slip | Dragging an auto-scheduled bar creates a start-no-earlier-than constraint. Remove it and let logic drive the date. |
| Links to summary tasks | Unpredictable dates and a broken critical path | Link the detailed tasks that actually depend on each other. |
| Manual tasks | Tasks that ignore changes around them | Switch to auto scheduled and add the missing logic. |
| Leveling left in place | Unexplained delays on tasks | Clear leveling, then resolve resource conflicts explicitly. |
| Percent complete with no actual start | Progress that the forecast cannot explain | Record actual start dates and remaining duration every period. |
| Very long tasks | Progress that is guessed rather than measured | Break work into tasks short enough to status with confidence. |
| Lags used as buffers | Hidden time nobody can see or manage | Model the waiting as a visible task with an owner, or hold explicit margin. |
What does a good monthly status cycle look like?
- Freeze the status date. Everyone reports progress as of the same day.
- Collect actuals from task owners. Actual start and finish dates, remaining duration and any new logic, captured by the people doing the work.
- Update and recalculate. Move incomplete work beyond the status date so nothing is forecast in the past.
- Run the structural checks. The fourteen DCMA checks plus the critical path test, every period, trended over time.
- Review the critical and near-critical paths. With the owners of the work on those paths, before anything is reported.
- Explain the variances. What moved, why, and what will be done about it.
- Publish and archive. The status file is saved, so every past forecast can be compared with what actually happened.
Which schedule health metrics should leadership see?
| Metric | How it is calculated | Common threshold |
|---|---|---|
| Baseline execution index | Tasks completed divided by tasks the baseline said should be complete | 0.95 or more |
| Critical path length index | Critical path length plus total float, divided by critical path length | 0.95 or more |
| Missed tasks | Tasks that should have finished by the status date but did not | 5% or less |
| Missing logic | Incomplete tasks without a predecessor or successor | 5% or less |
| High float | Incomplete tasks with more than 44 working days of float | 5% or less |
| Hard constraints | Tasks with constraints that override logic | 5% or less |
The thresholds are the commonly applied DCMA values, explained in full in our guide to the DCMA 14-point assessment. The trend matters more than any single reading: a baseline execution index drifting down for three months is an early warning long before the finish date moves.
How confident can we be in the finish date?
A deterministic schedule gives one date, and that date assumes every task finishes exactly on its estimate. A schedule risk analysis replaces single estimates with ranges for the most uncertain work and runs the network many times, usually with a Monte Carlo simulation. The result is a probability for each date: how likely the program is to finish by the contract date, and what date it could commit to with, for example, 80 percent confidence. GAO lists this as best practice eight, and it is one of the most persuasive things a program can show a customer, because it replaces optimism with evidence.
How is the schedule delivered to the government?
Delivery requirements come from the contract and its data requirements. On many Department of Defense contracts with earned value requirements, schedule data is delivered through the Integrated Program Management Data and Analysis Report, which calls for the native schedule file along with structured schedule performance data. The format, frequency and fields should be read directly from the contract early, because a schedule built without them in mind is painful to retrofit.
When is Microsoft Project the wrong tool?
Microsoft Project handles a single program schedule of several thousand tasks well. Enterprises running many interdependent programs with shared resources, or primes whose customers mandate a different tool, may be better served by enterprise scheduling platforms. The discipline described on this page applies in either case. What matters is that the tool, the delivery format and the earned value system are chosen together, at the front end, rather than discovered at the first submission.
How does Holly build the schedule on a Set Boundaries program?
On a new program, the integrated Microsoft Project schedule is not built alone at a desk and handed to the team. Holly builds it during the charter work, typically in the first one to two weeks, from conversations with the managers of each functional department and with the engineers, technicians and buyers who will actually do the work. Their durations, their dependencies and their knowledge of what can go wrong go into the file. The schedule is completed alongside the Project Charter, so the program starts with an executable plan rather than a list of target dates.
The purpose is visual. Anyone on the team should be able to open the schedule and answer three questions: where are we, where are we going, and what has to happen next to get there. A schedule that only the scheduler can read does not do that job.
What does a client see from week to week?
One schedule, used by everyone. The interdisciplinary team works from the same file every week, and a weekly two-week look-ahead keeps the critical path and the decisions it needs in front of everyone. There is no separate spreadsheet for engineering, another for purchasing and a slide deck for leadership. When the team, the customer update and the leadership review all come from the same source, the numbers agree and the discussion moves to what to do about them.
- The critical path is identified and kept in front of the team, so effort goes to the activities that decide the finish date.
- Near-critical work is watched closely, because it becomes critical quickly when something slips.
- Risk mitigation actions appear as tasks with owners and dates, not as notes in a separate register.
- Customer actions are scheduled too, so a late customer input is visible for what it is.
What happens when the critical path is threatened?
On many programs everything is a priority, which means nothing is. Holly's principle is to stop doing 30 percent of everything and start finishing the things that matter, and the critical path is how the team knows what matters. Anything that threatens it gets immediate attention. Holly calls that discussion "Critical Path Red": the stakeholders who can act see the impact and the options, and leadership makes an informed decision instead of discovering the slip at the next review. The schedule makes the tradeoff visible, so the program does not simply drift.
Holly prefers to build and maintain the schedule inside the client's own environment, so program information stays under the client's control and the team always has access to current status. When the engagement ends, the schedule and the habit of using it stay with the team.
How does Set Boundaries work on a schedule?
We start with a structural assessment of the existing file against the DCMA checks and the GAO best practices, and share the findings with the program team in plain language. From there the work is either repair or rebuild: fixing logic and constraints, restructuring durations with the people who own the work, loading resources, setting the baseline and establishing a status rhythm the team can sustain. For programs starting from scratch, this is part of front-end planning and the integrated master schedule build. For programs in trouble, it is the first step of program recovery.
If you are not sure the dates in your Microsoft Project file would survive a customer review, schedule a call and we can look at it together.
References
- US Government Accountability Office, Schedule Assessment Guide: Best Practices for Project Schedules, GAO-16-89G.
- Defense Contract Management Agency, EVMS program analysis guidance, the source of the 14-point schedule assessment.
- National Defense Industrial Association, Integrated Program Management Division, Planning and Scheduling Excellence Guide (PASEG).
- SAE International and the Electronic Industries Alliance, EIA-748, Earned Value Management Systems.
- US Department of Defense, Integrated Program Management Data and Analysis Report (IPMDAR) data item description.
Questions
Common questions
Can Microsoft Project produce a schedule that passes the DCMA 14-point assessment?
Yes. The assessment tests how a schedule is built, not which tool built it. Microsoft Project can produce a fully linked, lightly constrained, resource-loaded network that passes all fourteen checks. What usually fails is the way the file was built: manually scheduled tasks, constraints created by dragging bars, links to summary tasks and long tasks that cannot be statused. Those are habits, and they can be corrected.
Is Microsoft Project acceptable for an integrated master schedule on a defense contract?
For most programs, yes. Contracts specify what the schedule must contain and how it is delivered, not a brand of software. Many prime contractors and suppliers run their integrated master schedule in Microsoft Project Professional. Very large programs with many interdependent projects sometimes use enterprise tools such as Primavera P6, and schedules often have to exchange data with the prime or the customer, so the delivery format should be confirmed early.
Why do my Microsoft Project dates keep changing when nothing on the program has changed?
Usually because the schedule is reacting to something other than the work. Common causes are tasks left in manual mode, resource leveling that was run and never cleared, a moved status date, calendar exceptions, or links to summary tasks that behave unpredictably. A structural review finds these quickly. Once the network is clean, dates should only move when progress, logic or durations actually change.
How often should a program schedule be statused?
At least monthly, in step with earned value reporting, and weekly on programs that are in recovery or approaching a critical milestone. Each update should record actual start and finish dates, remaining durations and any logic changes, against a fixed status date. Updating less often lets small slips compound unnoticed, and updating without a status date produces forecasts in the past, which the DCMA assessment flags as invalid dates.
What is the difference between a baseline and the current schedule in Microsoft Project?
The baseline is a saved copy of the plan the program committed to. The current schedule is the live forecast that moves with progress. Microsoft Project stores up to eleven baselines and compares them with current dates to show variance. The baseline should only change through authorized change control; overwriting it to match the forecast removes the program's ability to measure its own performance.
Do you work in our files or build a new schedule?
Whichever serves the program. When an existing schedule is structurally sound, we repair and extend it so the team keeps its history. When it fails most of the structural checks, rebuilding the network with the people who do the work is usually faster than patching it, and the old file stays available for reference.
Is MS Project being phased out, and what is replacing it?
As of 2026, Microsoft has announced the retirement of Project Online, its cloud project and portfolio service, for September 2026, and points those users to Microsoft Planner, which has absorbed the features that were Project for the web. The Microsoft Project desktop client still exists and remains widely used for detailed scheduling. For defense programs, the real question is whether a tool supports logic, baselines and the DCMA 14-point checks. Set Boundaries continues to build and maintain integrated master schedules in the Microsoft Project desktop client.
What does FF FS SF SS mean?
They are the four task dependency types. FS, finish to start, means the next task cannot start until the first finishes, and it is the most common. SS, start to start, means the second task cannot start until the first starts. FF, finish to finish, means the second task cannot finish until the first finishes. SF, start to finish, means the second task cannot finish until the first starts, and it is rarely justified. Set Boundaries builds schedules mostly on finish to start logic, which the DCMA 14-point assessment checks for.
Have a question that is not here? Ask Holly on a short call
