2.4Cost estimating & budgeting

Early in a project, a PM faces the dilemma of having to estimate cost with a certain degree of accuracy despite limited time and information. Each estimating technique has situations where it excels, and "always use bottom-up for accuracy" is a naive, mistaken choice—performing a detailed bottom-up estimate during the early planning stage can waste effort. This section covers when to use each of the three estimating techniques and the difference in nature between the two types of reserve built into the budget.

2.4.1Choosing among estimating techniques

• Analogous estimating derives a cost from a similar past project's actual cost, adjusted for differences in scale and complexity. It requires little information and can be done quickly, but its accuracy is low, making it suited to rough-order estimates early in a project. Parametric estimating—function-point analysis, COCOMO, and similar—computes a cost from a statistical relationship or parameters. It is more accurate than analogous estimating, but presupposes reliable parameters and an accumulation of historical data.
• Bottom-up estimating produces individual estimates for each lowest-level item of the WBS (work package or activity) and sums them into an overall cost. It is the most accurate of the three techniques, but also requires the most effort and time to produce. It is appropriate once the WBS has been sufficiently detailed, in the planning stage and beyond.

2.4.2The cost baseline and reserves

• The cost baseline is the approved, time-phased project budget (the sum of per-activity cost estimates plus contingency reserve), serving as the reference against which actual cost (AC) and earned value (EV) are compared. Contingency reserve is a reserve built into the cost baseline to address known (identified) risks recorded in the risk register; the PM can draw on it at their discretion.
• Management reserve is a reserve for unknown risks (unforeseen events outside the project scope); it is not included in the cost baseline and appears only in the total project budget (cost baseline plus management reserve). Using management reserve entails changing the cost baseline, so it typically requires approval from higher governance (such as the sponsor), not just the PM's discretion.

A PM building the cost estimate for a new cloud-migration project uses a different technique at each stage. During the concept-approval stage, with detailed requirements not yet settled, the PM derives a rough budget (in the tens-of-millions-of-yen range) using analogous estimating, basing it on the actual costs of three similarly sized past on-premises-to-cloud migrations, adjusted for this system's scale differences, and uses it in the proposal to management. Moving into the planning stage, once the WBS is detailed, the estimate is updated to a more accurate one using parametric estimating (a model based on historical cloud-migration-effort data) that applies a statistical formula to the number of servers and volume of data being migrated. In the final stage of detailed planning, the PM gathers estimates from each responsible team for the individual migration tasks at the WBS's lowest level (per server, per application) and sums them via bottom-up estimating to finalize the cost baseline. Along the way, the risk register identifies a known risk—"three specific legacy servers may hit OS-compatibility issues (medium probability)"—so the PM builds a 2-million-yen contingency reserve into the cost baseline to cover that response cost, recorded as a reserve usable at the PM's discretion. Meanwhile, to prepare for an unknown risk outside the project's scope—such as "a major outage on the cloud vendor's side upending the project's entire premise"—the PM records a separate 5-million-yen management reserve, not included in the cost baseline, in the total project budget, with its use requiring the sponsor's approval. In this way, the estimating technique shifts from rough to precise as the project progresses, and reserves are clearly distinguished by whether the risk is known or unknown and by whose approval is needed to use them.

2.4.3Section summary

• Analogous = fast but rough; parametric = presupposes data, more accurate; bottom-up = most accurate, most effort
• Contingency reserve = for known risks, included in the cost baseline, at the PM's discretion
• Management reserve = for unknown risks, outside the cost baseline, requires higher-level approval