I recently met a software professional who was evaluating a software product to be bought and used in their company. As the discussion progressed, I realized that most of the metrics that he talked about were about the software itself. Lines of code, response time, hardware requirements, everything seemed to be about the software itself. I asked if he had tried to find out how it helped them better themselves at what they were doing. The confidence suddenly melted and most of the answer was hesitant.
ROI of a piece of software will never be about the software itself. It will be about you and how it helps you do things better. Neither the scale nor the metric will come from the software, it will come from you, your tasks and your processes. A piece of software might carry some features, but it is important to find out how valuable they are to you.
The easiest way to find out if the software will help or not is to look at your current constraints and problems. The software should either minimize a constraint or break one for maximum benefit. The tough part is to measure this in numbers. But are numbers so important? I think this is a better way of evaluating software than using false easy-to-measure metrics. Look at the software and its usage as a solution, and for that first work on your problems, not the software.