Measuring progress and percent complete

Actual cost tells you what has been spent; it says nothing about what has been achieved. A crew that has used 60 % of its budget hours on a wall is either ahead, on target or in trouble depending on how much of the wall is finished. Measuring physical progress is therefore the step that turns a cost ledger into a control system. It is also the step most often done badly, by asking the foreman for a percentage and writing it down. This lesson sets out the standard methods for measuring percent complete, how to choose among them, how to roll progress up from detailed items to a control account and the whole job, and how measured progress converts into earned hours and earned value.

Why opinions are not enough

A foreman's estimate of percent complete is quick and often reasonable, but it is biased in predictable ways. Early in an activity, people tend to overestimate progress because the visible work goes quickly; the last 10 % of an activity, with its details, corrections and cleanup, often takes far longer than 10 % of the time. Under pressure to show good results, estimates also drift upward. The result is the familiar pattern of an activity that is '90 % complete' for weeks. Objective methods, agreed before work starts, reduce these biases because they tie credit to something that can be counted or inspected.

The standard methods

Earned value practice recognizes a small set of methods for measuring progress on a work item. Each gives credit by a different rule.

Units complete

Units complete is the most objective method when the work has one dominant unit of measure that can be counted. Progress equals quantity installed divided by the budget quantity. Its weakness is that not every unit takes the same effort: the first hundred meters of pipe in an open corridor go much faster than the last fifty in a congested plant room, so straight units complete can overstate progress early. Separating codes by area or difficulty, or weighting units, reduces the problem.

Incremental milestones and weighted steps

For work that passes through distinct stages, the activity is divided into steps, each given a weight that reflects its share of the effort. Credit for a step is earned only when the step is complete, or in some systems partly as the step progresses. The weights are set before work starts, usually from the estimate or from historical data, and should not be changed to improve the numbers.

# Weighted-step progress for six wall pours on level 1
steps = {"layout": 0.05, "form one side": 0.30, "close and brace": 0.25,
         "pour support": 0.10, "strip and clean": 0.30}
walls = {   # steps completed on each wall at month-end
    "W1": ["layout", "form one side", "close and brace", "pour support", "strip and clean"],
    "W2": ["layout", "form one side", "close and brace", "pour support", "strip and clean"],
    "W3": ["layout", "form one side", "close and brace", "pour support"],
    "W4": ["layout", "form one side", "close and brace"],
    "W5": ["layout", "form one side"],
    "W6": [],
}
budget_hours = {"W1": 96, "W2": 96, "W3": 120, "W4": 120, "W5": 84, "W6": 84}   # bigger walls carry more hours
earned = 0
for w, done in walls.items():
    pct = sum(steps[s] for s in done)
    earned += pct * budget_hours[w]
    print(f"{w}: {pct:4.0%} of {budget_hours[w]} h")
total = sum(budget_hours.values())
print(f"Earned hours {earned:.0f} of {total}; activity {earned / total:.1%} complete")

W1: 100% of 96 h
W2: 100% of 96 h
W3:  70% of 120 h
W4:  60% of 120 h
W5:  35% of 84 h
W6:   0% of 84 h
Earned hours 377 of 600; activity 62.9% complete

Because each wall carries its own budget hours, the activity's percent complete is a weighted average, not a simple average of the six wall percentages. A simple average would give equal weight to the small walls and the large ones.

Start and finish methods

For activities short enough to start and finish within one or two reporting periods, measuring intermediate progress is not worth the effort. The 0/100 method gives no credit until the activity is complete; 50/50 gives half on start and half on finish; 20/80 and similar splits are also used. These methods are conservative (0/100) or slightly generous (50/50) within a period, but errors cancel out quickly because the activities are short. They should not be used on long activities, where they give badly distorted pictures for months.

Level of effort and cost ratio

Some costs produce no measurable output. A site superintendent's salary, a site office rental or a safety coordinator's time support the whole job. Level of effort (LOE) earns value automatically as time passes, so earned value always equals planned value and the schedule variance is always zero. That is correct for support work, but LOE should be kept to genuinely supportive items, because it masks problems: if LOE makes up a large share of a control account, the account will look on schedule even when the discrete work is behind.

The cost ratio method earns value in proportion to cost spent against forecast cost. It is acceptable for some services, but it simply assumes performance is on target, so it can never show a cost variance. Neither LOE nor cost ratio should be used for self-performed construction work that can be counted.

Rules of credit

Whatever methods are chosen, they should be written down before work starts as rules of credit: for each type of work, the method, the steps and weights, and who certifies the quantities. Typical rules also cover materials: whether value is earned when materials are delivered to site or only when installed. For cost control the usual practice is to earn value for installed work only, while the owner payment may include materials stored on site under the contract's conditions (Lesson 11).

Rolling up progress

Percent complete at the level of a control account, a trade or the whole project is the budget-weighted average of the items below it. The weight can be dollars or hours. For self-performed labor, budget hours are the usual weight; for the whole project, budget dollars. The calculation is simple, but it is often done wrong by averaging percentages without weights or by mixing items measured in different units.

# Roll up Riverbend structure progress by budget dollars (amounts $)
items = [  # item, budget, percent complete
    ("Footings and foundation walls", 612_000, 1.00),
    ("Slab on grade", 312_400, 1.00),
    ("Columns L1-L4", 284_000, 0.55),
    ("Elevated slab L2", 498_000, 1.00),
    ("Elevated slab L3", 498_000, 0.40),
    ("Elevated slab L4", 498_000, 0.00),
    ("Roof slab and parapets", 458_000, 0.00),
]
budget = sum(b for _, b, _ in items)
earned = sum(b * p for _, b, p in items)
simple = sum(p for *_, p in items) / len(items)
print(f"Budget {budget:,}  earned {earned:,.0f}")
print(f"Weighted percent complete {earned / budget:.1%}; unweighted average {simple:.1%}")

Budget 3,160,400  earned 1,777,800
Weighted percent complete 56.3%; unweighted average 56.4%

The weighted figure is the right one. Here the unweighted average is close, but on jobs where items differ greatly in size the two can diverge by many points.

Earned hours

For self-performed work, progress is often expressed as earned hours: the budget hours for the work actually done. Earned hours equal percent complete multiplied by budget hours. Comparing earned hours with actual hours gives the productivity factor that Lesson 5 develops. Earned hours are also easy for foremen to understand: 'you have earned 520 hours and spent 610' is clearer than a dollar variance that mixes wages and productivity.

Auditing progress

Because so much depends on percent complete, it should be checked. A cost engineer or project manager should walk the site with the superintendent at month-end for the main items, compare claimed quantities with drawings or models, and look for warning signs: items stuck at a high percentage for several periods, percentages that jump without matching hours, or earned value on work that has not been inspected. BIM quantities and reality capture (CON-260 Lessons 6 and 8) can make the check faster, by comparing installed elements in a scan or a progress-tracked model with the budget quantities.