Making software is science. I'm not talking about overengineering, just doing things right (with a minimum of automated testing, abstraction, architecture easy to modify, stuff like that). If you don't wan't to invest money in science, but only in business, get external providers for parts of your product, if not all of it. Stop making custom stuff that already exists, unless you can make something better (because it most probably ain't be cheaper, regardless the quality level).