What the schedule is for
A schedule has one job: to tell you, early enough to matter, whether the plan you are executing still resembles the plan you approved. Everything else — the presentation, the level of detail, the software — is in service of that.
Most schedules fail at this not because the logic is wrong but because nobody executing the work recognises it. A network built in isolation by a planner, handed to a site team who have their own sequence in mind, produces a forecast that diverges from reality within two periods and never recovers. I build the network with the people who will deliver it, because a plan nobody owns produces a forecast nobody acts on.
How I build it
Detail matched to phase. A twelve-thousand-line network at concept stage is not rigour, it is noise. Early schedules should be summary-level with honest uncertainty attached. Detail arrives as the scope firms up, and each increase in detail is a deliberate decision rather than a by-product of someone opening the software.
Durations traceable to a rate. Every duration on the critical path should be defensible: this many units, at this productivity, with this crew. If a duration cannot be traced back to a rate somebody is willing to stand behind, it is an estimate of confidence rather than of time.
Logic that reflects the site, not the org chart. Constraints and lags used to force a date are the most common way a schedule stops being useful. If the network needs a constraint to hit the required date, the honest output is that the required date is not achievable on current logic.
Float understood, not just calculated. Total float on a network with heavy constraints is close to meaningless. I look at where float actually exists in the delivery sequence and who is consuming it, which is often not who the report says.
Deliverables
- Baseline schedule with a written basis — assumptions, exclusions, calendar and productivity rates in one document
- Critical and near-critical path analysis, with the near-critical paths named rather than buried
- Progress measurement system, defined before progress starts being claimed
- Recovery and acceleration options, each with its cost and its risk stated
- Schedule quality review against DCMA-style checks, where a funder or client requires it
What good looks like six months in
The site team uses the schedule to plan their week. The project director can name the three activities that will determine the completion date. And when a date moves, everyone finds out in the period it moved rather than the period it became undeniable.