Can we use Occam's Razor and assume that nobody knows what would be the optimal tariff rates and if you don't have a reliable mathematical model the only choice you are left with is experimentation and A/B tests.