Why Progress Measurement Needs a Defined Basis

Article

Every progress report implies a measurement system, even when nobody has written it down. Someone decided that a drawing is 50 percent complete, that a foundation is “done,” or that earned value this month is a round number. If those decisions are not consistent from period to period, the S-curve will still plot. It will not mean the same thing twice.

What a basis of progress is

A defined basis states, for each class of work, the unit being counted, the evidence required to claim a quantity, and the rules for partial credit. Concrete might be cubic metres placed and accepted. Engineering might be deliverables issued at named gates. Installation might be tagged quantities that have passed inspection. The point is not that one unit is universally correct. The point is that the project picks one and keeps it.

The basis also states the data date, the treatment of out-of-sequence work, and whether remaining duration is independently estimated or back-calculated from percent complete. Mixing those methods without saying so is a common way for a programme and a cost report to disagree while both claiming to show “progress.”

What goes wrong without it

• Percent complete is adjusted to match a target curve, then treated as an observation.
• One period uses quantities; the next uses judgement; the variance is reported as a productivity event.
• Activities are marked complete while punch lists, testing, or as-builts remain, so remaining risk disappears from the model.
• Earned value uses a different breakdown from the schedule, so SPI and the critical path describe different jobs.

Keeping the basis light enough to use

A basis that nobody can apply in the time available will be ignored. It should follow the records the project already produces: inspection reports, quantity books, document transmittals, timesheets. Inventing a parallel measurement universe for the controls team is how reports drift from the site.

Rules of credit should be few and testable. “Design 30 percent when the model is issued for coordination” is usable. “Design 30 percent when it feels 30 percent” is not. Where judgement is unavoidable — for example, a unique commissioning sequence — name the person who signs the assessment and keep the same person, or the same checklist, from month to month.

What the basis does not replace

A defined basis does not make the forecast true. It makes this period comparable with last period. Forecasts still need remaining duration, logic, and a view of risk. But without comparable progress, those later steps are built on a moving definition of “done.” That is an avoidable weakness, and it is cheaper to fix in the first reporting cycle than in a dispute file two years later.