This forecast must be adjusted to reflect uncertainty. This "uncertainty factor" is derived through a sensitivity analysis that applies historical or benchmark data. The application of the results of this analysis to each predicted benefit provides an expected range of results, generally represented by the use of the best case, most likely case, and worst case results. The outcome is a range representing an objective quantification of the economic impact of the benefit.
In the same manner, analysts should provide for causation. For instance, there is a more direct correlation between the printer sharing and the number of printers required than there is between printer sharing and the revenue resulting from entry into new markets. Moreover, external factors, such as the economy, competitor actions (e.g. a new and much better product), or changes in fashion, can have a significant, negative (or positive) impact on the forecast. As this gap between cause and effect widens, so does the potential for inaccuracy in the benefit's forecasted effect. It is prudent, then, to make a final, conservative adjustment to allow for this risk.
When complete, these analyses of individual benefits are rolled up to produce a forecast of the economic effect of the component. This forecast also shows the financial risk of not making an investment in the component.
Taxonomy of Risk
Components of technology portfolios face risks internal to the organization. These can be organized into four categories: people, process, technology and project:
* People — a reflection of the competency of the personnel in the application and platform in which the initiative will be developed; their ability to effectively manage the deployment, manage the technology itself and use the technology to effect technical or operational improvements. Highly competent personnel who have taken on similar projects in the past lower the risk.
* Process — refers to the number of business process that are involved and the extent to which they are being changed. Is the change transparent to users or will they have to be retrained? If an initiative changes a business process(es), then there are additional risk factors such as resistance to change, transition from one process to the other, etc. Also, the creation of a new business process brings with it unique issues.
* Technology — if the application is new to the organization, it has higher risk than if it is a replacement for an existing application. If the application is new to the world, then the risk is even higher. Similarly, an increasing level of risk occurs if the platform on which the application is to be implemented is either new to the organization or to the world.
* Project — a measure of the internal support for the initiative. For example, decide how well the initiative is staffed and budgeted, the magnitude of the undertaking relative to projects that have been successfully implemented within the organization, and the visibility or priority of the project. Projects that do not have an executive sponsor frequently fail.
Risk factors from these four categories, after being weighted and assembled as a whole, form a "taxonomy of risk," which is a companion analysis to the business case and used in conjunction with the business case in the portfolio comparison.
Monte Carlo Simulation
The business case identifies a range of possible outcomes; the Monte Carlo simulation applies the taxonomy of risk to assign probabilities to the values lying between the best- and worst-case forecasts (e.g. how probable is it that the worst case will occur versus the best case versus all the values in between).
Once these probabilities are determined, a number of simulations can be run:
* Business Case Simulation, in which the distribution of possible outcomes is found by allowing each and every benefit and cost in the business case to take on a randomly selected value between its worst-case and best-case values. The distribution of these random values is determined by the probability assigned to each of the variables used to quantify the benefits and costs.
* Benefit Simulation, in which the impact of one benefit is determined by allowing variables for that benefit to take on random values while all the other variables remain at their most likely values.
* Assumptions Sensitivity, which accounts for a single variable being used to forecast any number of benefits and costs. The impact of these individual assumptions is found by randomly setting values for one assumption while all others remain at their most likely value.