The critical path cannot show a delay nobody scheduled

A schedule is not a report. It is a model.

The distinction sounds academic until something slips. A report tells you what happened. A model tells you what happens next, and it can only do that if the thing being modelled is actually in it.

That is the whole subject of this piece, and there is a published standard for it that almost nobody in the owner community has read.

What a programme schedule is required to contain

The US Government Accountability Office publishes a Schedule Assessment Guide, free and running to 240 pages, setting out ten best practices for a reliable schedule. It was written so that auditors could assess one.

Its definition of an integrated master schedule is the part that matters most to an owner:

An IMS constitutes a program schedule that includes the entire required scope of effort, including the effort necessary from all government, contractor, and other key parties for a program's successful execution from start to finish.

The guide is explicit that this is deliberate. It notes that others use the term to mean the prime contractor's schedule alone, and says its own use of integrated implies the incorporation of all activities, the contractor's and the owner's, necessary to complete a programme.

The first of the ten best practices says the same thing in the plainest possible terms. The schedule should reflect all activities in the work breakdown structure, including activities both the owner and the contractors are to perform.

And responsibility is assigned. The guide places development and maintenance of the integrated schedule with the owner's programme management office, not with the contractor.

What most owners actually hold

Set that against normal practice. The document reviewed at the monthly meeting is the contractor's programme. It was produced by the contractor, it models the contractor's work, and it is submitted for the owner's comment.

The owner's own activities appear, if at all, as isolated milestones. Design approval. Permit issued. Site access granted. Free issue delivered. A decision taken. Each one a diamond on a bar chart with no predecessor, no successor, no duration and no resource.

Those are precisely the activities most likely to run late, because they sit with the party that has the least scheduling discipline applied to it, and they are the ones the model cannot see.

Why that is not a small omission

An incomplete network does not give you a slightly wrong answer. It gives you a confidently wrong one, and the guide says so directly.

Unless the schedule represents the entire scope of effort and that effort is correctly sequenced through network logic, the scheduling software will report an incorrect or invalid critical path. With no accurate critical path, management cannot focus on the activities that will damage the finish date if they slip.

Read what follows from that in an owner context. An unlinked milestone has no successor, so nothing downstream moves when it moves. The software cannot place it on the critical path, because the critical path is the longest chain of linked activities and an unlinked activity is not in any chain.

The consequence is not neutral. A schedule built only of contractor activities can only ever show contractor delay. The owner's late approval is invisible to the model, and remains invisible right up to the point where somebody assembles it retrospectively, at which stage it is no longer a management question.

The ten practices, as a checkable list

They are worth setting out in full, because they are a specification and an owner is entitled to ask for each one. The guide notes they are in no particular order and are not a sequence of steps.

• Capturing all activities, as defined in the work breakdown structure, including those the owner and the contractors are to perform.
• Sequencing all activities, logically linked, with a predecessor starting or finishing before its successor, and with date constraints and lags minimised and justified.
• Assigning resources to all activities, meaning labour, materials, facilities and equipment, whether they will be available when needed, and any constraints on funding or time.
• Establishing the duration of all activities, realistically, using the same rationale, historical data and assumptions used for cost estimating.
• Verifying that the schedule can be traced horizontally and vertically, so that lower level activities roll up to milestones and different teams work to the same expectations.
• Confirming that the critical path is valid, the path of longest duration, which determines the earliest completion date.
• Ensuring reasonable total float, since unreasonably high total float on an activity or a path indicates that schedule logic might be missing or invalid.
• Conducting a schedule risk analysis, incorporating risk data into a statistical simulation to predict the level of confidence in meeting the completion date.
• Updating the schedule using actual progress and logic, with the people doing the updating trained in critical path method scheduling.
• Maintaining a baseline schedule under configuration management, with a basis document justifying constraints, lags, long durations and any other unique features.

Two of those repay particular attention.

Total float is a diagnostic, not just a property. Where an activity or a whole path carries improbable float, the guide reads it as evidence that logic is missing. That gives an owner a test that requires no judgement about the work itself: sort the schedule by total float and ask about the extremes.

And the basis document is the artefact most often absent. A schedule without one is a set of dates with no recorded reasoning, which cannot be audited and cannot be defended when the dates are challenged.

Accuracy that is not there

One line in the guide is worth quoting for its precision about a failure mode that has no other name.

Activity owners must be able to explain the logic behind their resource estimates, the guide says, and if resources are without justification, management will lack confidence in the estimated durations and the schedule may falsely convey accuracy.

Falsely convey accuracy is exactly right. A schedule is a document of dates to the day, produced by software, presented in colour. Everything about its form asserts precision. Nothing about its form distinguishes a duration derived from historical productivity data from one somebody typed because it looked about right.

There is also a sequencing rule that is broken almost universally. Planning and scheduling are different processes, and the guide states that in no case should planning be concurrent with scheduling: the work and the strategy for executing it must be planned before activities can be scheduled. On most programmes the schedule is where the planning happens, which is why the logic is thin and the durations are round numbers.

Where this joins the money

The eighth practice is the bridge to cost, and it is the one that most often goes unbought.

A schedule risk analysis incorporates risk data into a statistical simulation to predict the level of confidence in meeting the completion date, to determine the contingency, or reserve of time, needed for a level of confidence, and to identify high-priority risks. The guide says programmes should include its results in constructing an executable baseline schedule.

Note that contingency there is measured in time. Most organisations hold a cost contingency and can tell you its value. Very few hold a schedule contingency, can say what confidence level the completion date represents, or could produce the analysis that set it.

A completion date with no stated confidence level is not a forecast. It is a target that has been written in the space where a forecast belongs.

What happens to that date once it is allowed to move

, read: Every programme performs well against a baseline it can moveEvery programme performs well against a baseline it can move

What to examine

• Ask whether the schedule contains the owner's activities. Count them. Then check whether they carry predecessors and successors, because an unlinked activity is decoration.
• Sort by total float and look at both ends. Negative float means the network is already impossible. Very high float usually means logic is missing.
• Ask for the basis document. If none exists, the constraints and lags in the schedule have no recorded justification, and constraints are where inconvenient logic goes to be hidden.
• Ask what confidence level the completion date carries, and for the analysis that produced it.
• Ask who maintains the integrated schedule. If the answer is the contractor, the owner is being told about its own programme by the party whose performance the programme measures.

---

Sources. US Government Accountability Office, GAO Schedule Assessment Guide: Best Practices for Project Schedules, GAO-16-89G, December 2015. The ten best practices are quoted from the Concepts section, the definition of an integrated master schedule and the footnote distinguishing it from a prime contractor schedule from the same section, the consequence of an incomplete network for the critical path from The Critical Path Method, and the passages on resource justification and on planning preceding scheduling from The Integrated Master Schedule and Planning, Scheduling, and the Scheduler respectively. ISO 21502:2020, Clause 7, and the PMBOK Guide, Eighth Edition, November 2025, are cited without a link because both are sold rather than published.