Is there quantitative evidence showing that processes adapted from manufacturing work when applied to software projects?

Question

Most management practices derived from manufacturing were developed in an environment where measurements to the nearest 1/1000^th are common place. Most software project strive to estimate production +/- 10%. Given the difficulty the software world has in even simply quantifying its size, it seems translating process from the manufacturing world to the IT world must be done with care.

Does any publicly available quantitative evidence exist that shows the impact of implementing manufacturing-style process on software project where it did not previously exist? I'm not picky about the process, could be Six Sigma or something I've never heard of as long as its source was originally manufacturing.

Answer 1

Dave - You might find this article useful:

http://www.hbs.edu/research/pdf/08-001.pdf

It's a study of lean production techniques as applied to the processes of a software development company in India. It is generally positive about the adoption of lean techniques, although it does feature multiple caveats as to how robust the findings are in a bigger sense.

Answer 2

Your question is very broad with the use of the word "process" and the word "manufacturing." Many of the lean practices and six sigma problem solving practices can be effectively applied to software. But, I would argue that since the goals are very different between manufacturing and software that the process should not be directly transfered.

If you buy that Agile is a lean recipe that uses small batch size, cadence and feedback to manage the software process, you can look at the QSMA study of Agile teams performance verus more traditional teams. The QSMA Agile Impact report is located on the Rally web site.

Leveraging from Don Reinertsen's Product Development Flow book, we know that manufacturing tries to drive down variability to find an optimal point of lowest cost production. But by driving down the variability in software development, we drive out innovation. By driving out innovation, you reduce the potential pay-off of the software efforts. In software development, you are looking for the biggest payoff and not the cheapest cost of production.

Answer 3

I would use care with this translation of manufacturing to other industries. The problem is that often in manufacturing the 'solutions' tend to the 'replace the widget/ reorganize the line' options. Outside manufacturing, widgets and lines are less common and you can't just replace a person like you can a widget.

That said, you can find lessons in the way work is sequenced or building standard processes.

So I advise proceeding with some caution but proceeding anyway.