Twitter post from Edward Chenard laying out 10 truths about data science. Also a note from the journal Nature about how lack of a statistically significant effect should not be confused with a lack of any effect.
Data science has to be built on a strong foundation of well-defined goals and solid measurements. Without these, you will never be able to do great things. You also need the ability to experiment within your organization, run tests, learn from these tests, and only then can you start moving beyond to automated optimization, machine learning, and prescriptive analytics.