Thesis work paper Limit Research and Development to 10%-20% of Sales

Davidson (Davidson, 2001) explains that software firms typically spend too much on the product features, because the firms themselves have “over-developed” the product. He cites a common programming joke to illustrate this:
“It you work on a program long enough, it will eventually send email (Davidson, 2001)”
Such an inside joke does however demonstrate the “My Baby” syndrome that can get software companies in to trouble. By spending too long on software development the firms resources and capital are squeezed.
Further, because product development is slowed, a consistent cash flow is denied. Davidson (Davidson, 2001) and Tarter (Tarter, 2001) both recommend that software turnarounds begin with strong version control, limiting research and development of products to between 10% and 20% of sales.
What’s more developers should ship only a limited set of customer benefits (e.g. 10 features) for each product version.