How to measure project risk

Let's say you are in the retail outlet business -- a clothing store, a hotel, drugstore.  You build two hundred stores a year and you need to measure the risk in your business.  In other words, you can't wait until your stores are built to see how you are doing.  You can't use "on time performance" of the last few stores to predict the on time performance of the next few.

What key metric should you monitor to gauge risk or likelihood that your stores will be ready for advertising, staffing, and celebration?

I'll suggest that one useful metric would be what I call V/D index -- the "variation / duration index."  It's related to a question about "the 95% forecast range for all key milestones."  It is calculated like this:

Presume 200 projects to open new retail stores.  The projects all start on January 1 and consists of only one milestone:  Project Completion.  All projects are expected to last 180 days.

1. Ask each project manager to log, for each project, their forecast dates for completion of each critical project milestone.  E.g., "Enter the date range for which you are 95% of actual completion of this milestone."  For example, the project manager might put in that he is 95% confident that the store will be ready between June 15 and July 10.
2. Count the number of days in the range.  In this example, the number is (7/10)-(6/15) = 25 days.
3. Divide the number of days by the number of days for the project.  (25 days) / (180 days) = 13.9 V/D Index
4. This gives you a measurement of risk normalized by project duration. 

For the project above, report the 13.9 V/D Index.  This can also be easily aggregated for a portfolio of projects.

How do you make this real?  Once you start measuring it, put a reward system in place.  Reward people who hit tighter forecast ranges and punish people who complete milestones outside the range.

More concerned about budget risk than schedule risk?  This same method can be altered into a V/B Index, a "Variation / Budget Index."