82 Section 4
When an organization is mature in using EVM, performance indices are compared against variation thresholds
derived from historical data of past projects, which work as control limits. When the performance indices vary
within the thresholds, then no corrective action is required. A variation beyond the control limit can be a warning
sign that corrective action is recommended to counter the negative performance or to exploit the positive
performance being observed. This approach, based on thresholds, can also be used to implement management
by exception.
4.4.1.4 POTENTIAL CAUSES OF VARIANCES
Once variances are identified and measured, it is important to diagnose the causes so that, when necessary, action
can be taken to either mitigate or enhance their influence over the project.
Effective management does not aim at the absence of variance. Instead, good management is about identifying and
measuring variance, understanding its causes, and acting upon them, when required. This is a continuous process
where negative variance is treated early to prevent propagation and ripple effects, and where positive variance is
exploited as opportunities to improve the project’s final performance.
The EVM metrics and indices help to diagnose variation by providing detailed visibility of where the variance is
occurring:
Causes for variances can be identified by using the varying levels of the WBS to identify the scope components
with the larger variances;
Variances can be calculated on specific segments of the project scope, for example:
Scope assigned to specific contractors or functional units,
Type of work (e.g., engineering, excavation, programming), and
Type of resource engaged in doing the work.
This type of top-down and segmented analysis is useful for understanding where, who, and what type of work is
overperforming or underperforming. This analysis focuses the action where it is most needed.
When an organization is mature in using EVM, correlations regarding the status of the project can be established
using EVM metrics, performance indices, and potential causes for variance based on historical data and lessons
learned. In other words, common symptoms can be consistently linked to potential causes, as shown in Table 4-2.