Systematic experimentation - the value of samples