Machine learning in the Oil & Gas industry has met with varied success and no shortage of hype. Barriers to success in the industry include data challenges, inflated expectations, and a lack of trust from those who stand to benefit from its results. In this presentation, we will be showing the bulk of our Spirit River Study from the SPE Subsurface Workshop, plus a whole lot more. We will explore trust-building approaches that include feature selection, interpretive visualizations, stability considerations, “What’s a good R2?”, and a Man vs Machine simulation. In the simulation, we compare traditional type-curve generation approaches with those based on machine learning models, using data from the Marcellus.