Project Estimating Best Practices

Leverage history to better predict the future. Measuring and recording project tasks provides an empirical basis for estimating those tasks in the future. Professional Services Automation (PSA) software can retrieve historical similar projects, replicate the relevant portions by categories or work breakdown structure (WBS), permit adjustable parameters based on project specifics, apply a 'complexity factor' based on project uniqueness or risk, and produce a template or draft estimate for further refining.

If you have a history of accurate project estimates, memorialize those estimates as templates in your PSA software and categorize them by project criteria or metadata so you can automatically retrieve and repurpose the most relevant data to create new project estimates. This will accelerate the estimating process and apply previous lessons learned.

If you do not yet have a history of accurate estimates, or for first of a kind (FOAK) projects, identifying an accurate estimating technique is the first order of business. There are a couple of approaches to do this. You can apply a single estimating method but create multiple estimates based on varying forecast measures. This will result in a range of estimates, such as Low, Moderate and High estimates. Actual project results compared to each forecast measure will show which estimating method was most accurate.

This approach is a spinoff of the Program Evaluation and Review Technique (PERT), which calculates a weighted average of the most pessimistic, most optimistic and most likely case. For a PERT calculation the most likely case receives a higher weighting and the final result will bend toward the side (optimistic or pessimistic) which is furthest from the most likely.

An alternative technique is to apply different estimating methods to the same project to see which most approximates reality. We've done this a few times and found that comparing the Analogous (i.e. top-down), Analytical (i.e. bottom-up) and Delphi estimating methods tend to flush out the most significant differences that can be analyzed and compared to actual results.

Analogous estimating, often called top-down planning, heavily leverages the experience of your most seasoned executives. This technique is most effective when the service provider repeats similar projects. This method still leverages historical project data to create a new estimate, but the estimator may apply judgment to modify estimating factors based on the client or project specifics.

Analytical estimating, often called bottom-up, works best for projects with many variables or FOAK projects. A WBS is needed and the estimator may need to extract, group or parse work packages of lower level grouped tasks. These estimates are the most time consuming and complex but also the most extensible and generally provide the most accurate project budgets for unique or complex projects. The estimating time can be accelerated by starting with dynamic templates. The biggest challenge with this method is estimators that pad their estimates to cover potential overruns. It's a pervasive problem that defeats the process and result.

Parametric estimating uses an underlying unit of work to measure and extrapolate a financial estimate. For example, a builder may use square feet, or an accountant may use the number of forms. The estimator will determine the number of units of work for the project and multiply by the cost per unit. This estimating method is generally limited to projects with standardized cost units and a high degree of repeatability.

Not sure what method will work best for you? With some types of projects that have proven difficult to accurately estimate, we have found comparing estimating methods can do the trick.  Applying multiple estimating techniques is a significant investment, however, determining the most accurate method is a critical success factor that will generate a payback on the investment multiple times over.

The Delphi technique is designed for this purpose. It uses expert consensus to compare, contrast and refine different estimating methods until the best method is agreed upon. Multiple experts create individual and anonymous estimates with supporting rationale which are then circulated back to each estimator for further consideration and updates. Consensus is generally achieved after a few iterative cycles.

Regardless of method, technology is the facilitator. Several PSA systems permit multiple estimates and budgets to be applied to each project so actual results and variances can be compared among different estimating methods during the course of the project. You can also drill down on the factors that separate the estimates to analyze their impact.

PSA systems can also include estimating templates with checklists, tasks and alerts for possible omissions. Using project templates in this way permits tasks to automatically insert subtasks, precursors or dependent conditions. For example, when a quality control task is added, the system automatically appends a subtask for time to resolve the inevitable defects or remediation.

The template approach aids the estimator in identifying scope and effort down to a task level and can be updated with actual results that are then applied back to the original template.