The question is often asked: "How do I calculate the automated testing Return on Investment (ROI)?" Where I work, we have come up with an extensive spreadsheet that allows us to calculate this ROI, but wanted to let you know about a very basic and good starting point ROI calculator example provided by IBM, called "IBM Rational Quality Management ROI calculator."
For more detail go here http://www-306.ibm.com/software/rati..._Rational.html
It's a good start which can be expanded for additional criteria and the calculations can be used for various tools, not just IBM/Rational.
Mercury also published a standardized automation ROI spreadsheet a few years ago. It was directed to their 3rd party vendors and didn't get full circulation like it should have. With the changes in management there it was probably lost in the shuffle.
I agree it is going to take some time, but since we are at the start of trying to document/figure out how we can start determining this information, then it is a great start for me [img]/images/graemlins/smile.gif[/img]
This is always a tough one to predict and until you have some hard data (experience) to analyze and report on you cannot really determine ROI. That has been my experience.
I see some of these 'tools' from the Vendor's as another "Automagic" smoke & mirror. I know, I've been in consulting and have used both Rational and HP/Merc tools in the last 10 years. I always have tried to use a real world experience/project to explain what can be done with automation if properly planned, implemented and managed.
These 'tools' tend to not factor in the learning curve (time and money for initial learning & setup) that all automation projects incur. This is the real sticker shock that people do not realize or like. And this is part of the reason initial implementations fail miserably.
But, thanks for posting this Elfriede. To me it is something to use as an example of the potential of automation from an economic standpoint. I firmly believe automation is a good tool to aid in testing. In this instance Regression Testing is something that is time & resource consuming and if you can get a bunch of machines to do the work load you can get economies of scale for your efforts. Machines are cheaper to use and can run 24x7 if you set it up that way, people cannot do this.
What I would like to see sometime (if someone can find an Excel worksheet) is a model of ROI for a new software system for a business. To me this would more accurately mimic what will be incurred for automation.