How much of what we call data science is just clever data storytelling rather than rigorous statistical inference? I've been working with a retail dataset for months, and the more I dig into the numbers, the more I realize how easily p-hacking and confirmation bias can slip into even well-intentioned analyses. We teach p-values and confidence intervals, but in practice, decisions often hinge on which model the stakeholder prefers. When does a dataset actually speak for itself, or are we always projecting our own narratives onto it?