2.1Activity definition & sequencing

The first step for a project manager (PM) building a schedule is to further decompose the work packages from the WBS into activities at a granularity where start and completion can actually be tracked, and to correctly define the dependencies among them. Getting these dependencies wrong leads directly to real-world failures downstream—sequencing work that could have run in parallel and squeezing the deadline, or the reverse, running dependent work in parallel and triggering rework. This section covers the four types of precedence relationship, lead and lag, and three-point estimating, which improves the accuracy of duration estimates.

2.1.1Precedence relationships (FS/SS/FF/SF)

• FS (Finish-to-Start) means the successor can start only after the predecessor finishes. It is the most common relationship, used where the order is physically or logically fixed—for example, "painting can start only after the wall's base construction finishes."
• SS (Start-to-Start) allows the successor to start at the same time as the predecessor starts. It is used where work can begin without waiting for full completion—for example, "preparation for basic design can start in parallel as soon as requirements definition begins." FF (Finish-to-Finish) coordinates both activities to finish at the same time. SF (Start-to-Finish) has the successor finish when the predecessor starts—the least common relationship in practice (such as a shift handover).

2.1.2Lead and lag, three-point estimating

• Lead is the amount of time by which a successor can be pulled earlier relative to the precedence relationship (e.g., under an FS relationship, if the successor can start 2 days before the predecessor finishes, the lead is 2 days). Lag is additional waiting time inserted after the precedence relationship (e.g., after concrete pouring finishes, if the successor is pushed 3 days later for curing, the lag is 3 days). Do not confuse the two: lead shortens the schedule, lag delays it.
• Three-point estimating derives a duration from an optimistic (O), most-likely (M), and pessimistic (P) value. The triangular-distribution expected value is (O+M+P)/3, while the PERT weighted average, which better approximates the underlying uncertainty, uses (O+4M+P)/6, giving a risk-informed range rather than a single-point estimate.

A PM is building the schedule for a core-system renewal project. In principle, "integration testing" starts only after "unit testing" finishes (an FS relationship), but with the schedule tight, the PM is considering whether the environment-setup work for integration testing—just for the modules whose unit tests finish first—can start 2 days earlier. This corresponds to setting a lead (2 days) that pulls the successor's start earlier while preserving the FS relationship. However, the PM also wants to preserve the constraint that "full-scale integration test execution begins only once unit testing is complete for all modules," so the lead is applied only to the pullable sub-task of "environment setup," not to test execution itself. Separately, another activity requires a lag (1 day) after "hardware delivery from the outside vendor" finishes, to allow the equipment to acclimate to room temperature before installation begins. If this lag were overlooked and the schedule had installation start the same day as delivery, the risk of equipment defects or faulty installation right after delivery would rise—so the lag must be treated not as wasted time to be trimmed, but as time deliberately reserved to protect quality and safety. Furthermore, for a new module's development duration, the PM gathers three estimates from the assignee—"5 days if things go smoothly, 8 days normally, 20 days if trouble occurs"—and, given the high uncertainty of new development, decides to compute the estimate using the PERT weighted average. Rather than a simple average (triangular distribution, (5+8+20)/3 = 11.0 days), adopting the PERT weighted average, which weights the most-likely value ((5+4×8+20)/6 = 9.5 days), yields an estimate closer to practical judgment, not overly pulled by extreme optimistic or pessimistic values, and this is what gets built into the baseline.

2.1.3Section summary

• Precedence relationships come in 4 types: FS (most common), SS, FF, SF. Lead pulls earlier; lag adds waiting time—opposite directions
• A lag may be time deliberately reserved for quality or safety, and is not always something to trim
• Three-point estimating uses the PERT weighted average (O+4M+P)/6 to derive a practical duration not skewed by extreme values