The building blocks of Estimation
Estimation is about predicting an uncertain quantity in advance. In project management, we usually estimate for resources, duration, money, effort for an activity / project. Accuracy of estimation is critical for a project success: Underestimation causes loss to the performing organization whereas over-estimation might cause business / goodwill loss with the customer.
|Accuracy category||Expected accuracy|
|Rough Order Estimate (ROM)||-25% to +75%|
|Definitive Estimate||-5% to 10%|
During pre-sales, we might follow ROM and subsequent re-estimations are done with tighter accuracy. Although every industry follows different estimation model(s), they use one or more of below estimation building blocks:
1. Analogous (TOP DOWN) estimation: Is a gross value estimating approach with some adjustments for known differences. It uses historical data about similar activity/project. This approach does not provide accurate estimation but require only little time, project information and technical expertise.
Example: To estimate for similar construction project undertaken 3 year back, we could take actual cost of construction of previous building and then adjust for increase in inflation or construction cost.
2. BOTTOM UP estimation: Keep decomposing (dividing) the ‘item to be estimated’ to sub-items in a nested way (like tree branch) until you reach a point where all leaf items can be estimated with better accuracy. Now estimate leaf items and aggregate for whole item. The below picture illustrates this. This approach takes more time, require detailed project information and technical expertise but provides accurate estimation.
Example: To estimate total cost of construction of a big building, estimate the components like foundation, each rooms, lift, common area, bore well, car parking, security bay, sump, swimming pool... and sum it up to get the total estimate.
3. Parametric estimation: Uses RATEs from historical data to estimate.
Example: To estimate total cost of construction of a flat, apply PRICE PER SQ FT * AREA in SQ FT.
4. Three point (PERT) estimation: Do three estimates: Most likely (m) estimate, Optimistic (o) estimate (best case) and Pessimistic (p) estimate (worst case). Usually Beta Distribution is assumed and mean estimate and standard deviation are found by below formulas. This concept is taken from PERT (Program Evaluation and Review Technique).
- Estimate = (o+4m+p) / 6
- Standard Deviation = (p-o) / 6 (note that most likely estimate is not used in this formula)
Example: If it is estimated that a job most likely to take 6 days, can be completed in 4 days in best case and might take 10 days in worst case. Then
Estimate = (4 + 4 * 6 + 10) / 6 = (4+24+10) / 6 = 38/6 = 6.33 days
Standard Deviation = (10-4) / 6 = 6/6 = 1 day 8 6 2 3
19 4 9 5 7