The revenue-maximizing price is the price that gives the best tradeoff between selling more units and earning more per sale. It sits at the sweet spot where expected revenue is highest: probability of buying times price. section 2.11 section 2.11
A simple way to use it is to test candidate prices and compare the model’s predicted purchase probability at each one. If a shirt sells less often at $40 than at $20, the higher price only wins if the extra margin makes up for the drop in sales. section 2.11 section 2.11
The catch is that this can go wrong when the model hallucinates how buying probability changes with price. The evidence here warns that the probability may rise, jump around, or stay unrealistic when price is varied. section 2.11
