How do we select the best possible alternative, when there are still a large number of unknowns?

Approaching decisions armed with adequate information is, of course, the ideal, but it's not always possible given the time constraints of a project. Nor do we always have the leisure to address each critical issue that factors in a decision. So we rely on time-tested tools such as decision analysis to help us make the best possible decision, given the information at hand. Decision analysis and decision trees are most often helpful when they're closely tied to hard financial measurements, but they can also help us make good choices when some of the factors are unknown.

When you must choose among several options, decision analysis can help you identify the tradeoffs of each choice. Another approach to decision-making when some of the factors are unknown is to analyze the impact if the risks are realized, and translate the risks into hard numbers, such as dollars or lost time. For example, consider a team that must choose a solution from vendor A, B, or C. The vendors' proofs of concept are dead-on for some of the highest-priority requirements, but there's still a lot unknown about how each solution will perform when it's implemented. A table of the options and impacts can help guide the decision

	Vendor A	Vendor B	Vendor C
Current relationship with vendor	None – vendor is new to industry	Active – vendor has delivered other solution to our organization	Previous – vendor relationship ended due to high cost of ongoing maintenance
Cost to manage vendor relationship (annually, based on anticipated % of headcount/salary, estimated by exchanges with vendor).	$25,000	$10,000	$15,000
Cost of solution	$500,000	$600,000	$800,000
Cost of ongoing maintenance (annual)	$5,000	$6,000	$16,000
Ballpark cost to build an internal solution to fulfill requirements not met by the vendor's solution	$200,000	$125,000	$10,000
Cost to restore if the system fails	$25K for every hour of business lost.	$25K for every hour of business lost	$25K for every hour of business lost
Likelihood of quick system restore	Unknown – small vendor, probably has limited ability to respond quickly; contract says vendor will cover business losses after 8 hours.	Medium – vendor has proven that they can restore within 6 hours.	High – large vendor with multiple redundancy, failsafe, and backup options, cites 2-hour response
Cost of system failure, assumes 2 failures in 3 years ($25K per hour x expected hours of failure)	$200,000	$150,000	$50,000
Total cost over 3 years	$1,190,000	$1,073,000	$1,003,000

Drawing a grid can be a very useful exercise, even though, in this case, it's difficult to estimate the exact cost of an internal build, and it's impossible to know how often the system will fail. Still, the grid enables the team to calculate a rough dollar estimate of the total cost over three years. Depending on the risk tolerance of the organization, a grid can at least serve as an organizing aid for choosing a vendor.