Just to clarify: I should have said that you have more water in the TOTAL of all vessels than can possibly fit in the barrel. Each vessel is smaller than the barrel by itself. This is exactly the situation when you have more project ideas than you can fund. Which ones do yo choose first?
Before I give the final answer on my NPV (Net Present Value) vs. ROI (Return on Investment) question, I'd like to give an analogy that I invented in order to explain why the answer is what it is. As I wrote, even finance people can get this one backwards.
Imagine you have a large barrel. Your objective is to fill it with sugar water, and to have the most sugar in the barrel as you can.
You must fill the barrel from an assortment of different sized glasses and mugs, each with different concentrations of sugar within them. The CONCENTRATION of the sugar in each vessel is the ROI. It is just a ratio. The TOTAL sugar dissolved in each vessel is the NPV.
You have more water in the combined totals of vessels than can possibly fit into the barrel.
Do you pour in the most concentrated (highest ROI), or with the most absolute sugar (NPV)?
You goal is to maximize the sugar in the barrel (NPV). Does it also maximize ROI?
You are absolutely right about sunk costs. An organization should always be looking at incremental costs and incremental returns when prioritizing projects. A dilemma that faces high tech, is that it takes some investment to even forecast schedule, investment, and revenue. This can work for or against a project: A product that forecasts weaker sales than hoped may still be executed, because the incremental effort to complete the project is low. Conversely, a project may be cancelled even if there is already a good investment behind it, it the remaining investment does not justify the return.
The opportunity cost comment is true as well.In general, if you propoerly ranked the projects, you should be comparing a new idea against your best unfunded project.
The inspiration for my remark about 2 decimal places was a complaint by (IIRC) Norbert Wiener, who had accountants as "computers", (the term in those days meant people), to work on ballistics tables for artillery. He said something along the lines of "It didn't matter if the number in question was the calibre of the gun in millimetres, or the range of the shell in miles, they always worked to 2 decimal places".
(Find the missing spacecraft lurking in that quote. :-)* )
I'm lucky to be an engineer and to also have earned an MBA from a top school. A lot of what I learned in both disciplines is impractical, or "not how things are done in the real world."
For engineering managers, I always preach they know and understand two things: (1) Sunk costs, and (2) Opportunity costs.
For sunk costs, if you've ever heard someone say, "We have to keep going on this project. We've already spent $1M and 6 months...", then you know sunk cost. I still hear statements like this from people who are very intelligent. The issue is framing: Decisions based on cost need to be framed from now ("time zero"), not when expenses started accruing for a project.
The other key concept is opportunity cost. If you've ever heard an engineering manager say, "We can develop that in house because I have people who have done it before...", then then next thing you should say to her is, "But should you?" In this example, the opportunity cost is the difference between the value of the work the engineering manager says her team can do versus the other, perhaps higher value, work the team could do. Opportunity cost is the most important factor in make vs. buy decisions.
If an engineering manager can grok these two concepts, then her financial education is (mostly) complete :-)
Just as a comment, ROI can take into account time value of money. That is, future cash flows may be discounted before the ROI calculation. However, you are correct that ROI doesn't take into account the total size of the investment or revenue, and this is the big deviation from the NPV ordered list.
Quite agree that its nice tohave engineers with financials baiscs. As one grows up the ladder, you got to understand revenue part and if you have basics in finance it helps a lot. Even if you remain in technical side its hard to stay away from finance because everything boils down to numbers.
The problem with using ROI to select doing one project over another is that it doesn't account for the time value of money and doesn't account for the difference in total revenue between each project, or the timing of that revenue. If I must choose between doing a smaller project (lesser lifetime revenue) with a huge ROI or a much bigger project (greater lifetime revenue) but with a smaller ROI, then choosing the higher ROI project may ultimately mean I make less profit. The NPV of the larger project could be much higher -- especially if it's revenue stream starts earlier than that of the smaller project.