ScrumROI
From Epowiki
On the scrum email list Jeff Sutherland gave the following analysis for the ROI of scrum:
Lots of people what to know what the ROI is on Scrum. Let's give them one.
Here is an approach to laying out an ROI for Scrum. I've toned it down a little because the real ROI for properly implementing Scrum is too big to be believable to people living in the land of the Waterfall where projects are always on time even when developers know they are late.
First, get the best available independent consulting group to get concrete baseline data on all development teams. Determine function points already available in every product in all product lines, and determine the net deliverable incremental function points per developer per month.
I know developers don't like function points but ROI is for bean counters and they will like them.
Any company who hasn't done this level of analysis has either no data or very poor data on productivity. Of course that is most companies.
Insert Scrum company wide and get real data on incremental function points per developer per months for a few years. Then you have some real data to base an ROI on. This was done at IDX for a team that grew to 567 developers during the period 1996-2000. Not all of them had fully implemented Scrum as we had a bunch of COBOL programmers on a multi-million line clinical system that were somewhat impervious to change (but they saw the handwriting on the wall).
I'm laying out the stats as follows:
- Industry baseline is 3 function points per month for typical commercial software. That is actually what it was when I got to IDX.
- 10% of teams got to 20 function points per month.
- 5% of teams failed (actually I don't think it was this high). Industry standard team failure stats are 31% and they fail a lot later.
- 80% of teams got to 6 function points with Scrum.
- Assume 30% of the teams move to Scrum level the first year.
I leave the math as an exercise for the reader. The net gain in productivity is 173% the first year. Let's say you have a 60 person development team and you pay an average of $100K per month per developer fully burdened. You probably are already doing outsourcing to meet this number.
So the 60 person team used to deliver $6M of industry standard software. At the end of the first year they have delivered $10.4M work of industry standard software for a gain of $4.4M.
I think we should budget Scrum implementation for a 1000% ROI. This will be substantially higher than any other item on the CFO's list, even PatientKeeper's product which gets 400% ROI for healthcare institutions, probably the highest in the healthcare business.
Take 10% of $4.4M or $440,000 and spend that on implementing Scrum the first year for a 60 person team. That ought to be enough to get it going.
I think this is what Ken means when he talks about Extreme Business Value.
Jeff Sutherland CTO, PatientKeeper Inc.
