Weighted Percent Complete: Why Simple Averages Get Rollups Wrong

Most spreadsheet trackers roll up progress with AVERAGE(). That's fine exactly as long as every child item is the same size — which, on a facility project, is never. The result is a summary that overstates readiness whenever the small easy items finish first, which is always.

The worked example

Milestone 1
├─ Activity 1.1 (weight 3)
│   ├─ Task A (weight 1, 100%)
│   └─ Task B (weight 1,   0%)   → Activity 1.1 = 50%
└─ Activity 1.2 (weight 1)
    └─ Task C (weight 1,  80%)   → Activity 1.2 = 80%

Simple average:   (50 + 80) / 2           = 65%
Weighted rollup:  (3·50 + 1·80) / (3 + 1) = 57.5%

Activity 1.1 is three times the work of 1.2, and it's half done. The simple average hides that behind Task C's easy 80%. Seven and a half points may not sound like much — until it's the difference between “we start up Thursday” and “we don't”.

The formula

parent % = Σ(child.weight × child.%) / Σ(child.weight)

Applied bottom-up: tasks → activities → milestones → phases → project. Two properties matter in practice: only leaf tasks take a manual percent (everything above is computed, so nobody can hand-tune a summary), and weights default to 1 so the scheme degrades to a plain average until someone deliberately says otherwise.

Choosing weights without a fight

• Weight by rough effort share, not precision — 3-to-1 beats arguing about 3.2-to-1.
• Set weights when the tree is built, review them when scope changes, and record who changed what.
• Watch the edge cases: siblings that are all zero-weight, empty parents (“no work defined” is not “0% done”), and deleted items that must leave both the numerator and denominator.

Fieldcurve computes exactly this rollup transactionally on every edit — and every computed number can show its contribution breakdown (weight × percent, child by child), so when a manager asks “why is this milestone only 57.5%?”, the answer is a hover, not an afternoon.