Delivering the Work: Projects, Tasks, Time and Cost
A plan with real dependencies, people assigned to tasks, time booked against them, and cost accumulating as it is incurred rather than reconstructed at the end.
McKinsey's research on large projects puts average cost overruns in the region of 28 to 33 percent. Overruns of that size are not usually one catastrophe — they are a hundred small facts that nobody could see in time.
The plan is a copy, taken deliberately
When a work order is taken up for execution, QuantX Ops copies the estimate into a plan. From then on the plan is the live tree and the estimate stays as the baseline you priced against. Re-planning does not rewrite history, and "we are over because the scope changed" becomes a claim you can show rather than assert.
Cost sits on the leaves
Budget and hours belong on the tasks that do the work, not on the sections that contain them — a parent's figure is the sum of its children. This sounds pedantic until you try to measure progress: a section carrying its own cost double-counts, and the earned-value figures quietly become nonsense.
Three sources of cost, one total
| Source | How it arrives |
|---|---|
| Labour | Time booked against tasks, priced at the rate frozen when the person joined the project |
| Material | Stock issued from stores, at moving-average cost |
| Expenses | Approved claims and supplier bills tagged to the job |
The labour rate is frozen on assignment deliberately: a pay rise in March should not retrospectively change what January's work cost.
Schedule and progress
Tasks carry dependencies, so the critical path is computed rather than guessed. Progress against the baseline gives you schedule and cost performance — and, more usefully than any index, a list of the specific tasks that are late.
References
- Reinventing construction: a route to higher productivity — McKinsey Global Institute