Getting tough on ROI. InformationWeek has a good article that explores various methods that companies are using to evaluate the return on investment from IT projects. Among the companies with best practices for managing ROI:

1. Schneider National, Inc., which categories IT projects according to whether they reduce costs, increase revenue, or simplify business processes, allowing management to use a different set of criteria in justifying each category of project.
   
   
2. Schlumberger, which uses worst-case, likely case, and best-case scenarios, allowing management to focus on those factors that reduce the risk of realizing the worst-case.
   
   
3. Vanguard, where executives meet in "sunlight sessions" to debate and challenge the projected benefits of a proposed IT initiative, allowing overly optimistic assumptions to be caught before the project is approved.
   

The article also includes a case study of Citibank, which used a complex valuation simulation tool to trace the potential impact of an executive portal project on Citibank's stock price. Although such an approach has certain appeal, I believe that the complexity far outweighs the benefits for all but the largest companies. In my experience, most companies would benefit simply by doing a reasonable cost/benefit calculation at project initiation, verifying the assumptions, managing risks, and measuring the results after implementation.