Your AI Subscription Pricing Is Losing Money on the Customers You Care About Most

Do you actually know which of your AI customers are making you money and which are quietly destroying your gross margin? In episode #375, Ben Murray breaks down the shape of AI usage and why the distribution curve of your customers determines whether your AI subscription product is profitable. This is why Anthropic and GitHub changed their pricing. Heavy users on a flat subscription can quietly turn a 40% gross margin into a negative one, and most finance teams are not tracking token usage by customer in enough detail to see it coming. The three AI usage distribution scenarios every SaaS CFO needs to model: normal, right skew, and left skew, and what each does to your gross margin Why a right-skewed distribution means your light users are subsidizing your heavy users, and how to spot when that subsidy stops working How a left-skewed distribution can leave 80% of customers unprofitable and drag overall gross margin into the negatives Why median, mean, and P90 token usage by customer are now core SaaS finance metrics, not just product analytics What finance needs from product and engineering — usage by customer, model mix, input and output token pricing — to run real pricing scenario analysis Tune in before your next pricing review and find out where your AI margin is actually leaking. Resources Mentioned Ben's AI metrics course with the usage distribution template and free preview: https://www.thesaasacademy.com/ai-finance-metrics-saas AI readiness quiz: https://www.thesaasacademy.com/ai-finance-metrics-saas