During March Madness, it seems like anything can happen. A number one seed can take the whole tournament or be upset by a low-ranking underdog. You can predict, with your bracket, how you think it'll turn out, but only the lucky few will have intact brackets by the Final Four. Maybe you pick based on season-long data and NBA prospects. Or maybe you pick based on school colors and mascots.
With as much time as you might spend developing the perfect March Madness bracket, do you spend that much time on your "marketing bracket"? (That's not really a thing, but go with me here.) There's a marketing term called split testing (or A/B testing) and it's very similar to the scientific method or NCAA brackets: make predictions based on past data, see how it turns out, and apply what you've learned next time.