Field notes · Pricing & monetization
Usage-based pricing is the most frictionless front door software has ever shipped. The brakes for the month it runs away from someone are the part nobody builds.
What I've come to believe
The smartest idea in software pricing right now is to treat pricing like a product: something you ship, instrument, and improve, not a one-way door you open once a decade. But every product has a back end, and the back end of usage-based pricing is the month spend runs away from someone. We are building a beautiful front door to consumption and almost no brakes for when it goes wrong.
I sat in on a talk at Stripe Sessions this year that I have not stopped thinking about: Scott Woody, who built Metronome and now runs billing at Stripe, in conversation with Shaa Alagumuthu, who leads monetization at Anthropic. The frame is sharp and correct. Pricing is a product. AI is accelerating value in a non-linear way while most companies still price like it is 2010, seat-based and frozen. So cannibalize yourself before a competitor does, treat pricing changes as reps in a gym instead of one-way doors, and put one accountable person in charge of shipping them. I agree with almost all of it. I just read it from the other end, the way I read everything.
Usage-based pricing solves a real problem: it aligns the bill with the value delivered, so every dollar a customer spends is a dollar of value they captured. That alignment is the genuine breakthrough, and the talk is right that you cannot pull it off without extreme instrumentation. You have to show a customer their spend at every level, the organization, the department, the individual, the agent, the single workflow, or they will not trust the bill.
But look closely at what all that instrumentation is pointed at. It is pointed at consumption: where is spend happening, what is working, where can the customer do more of it. The same mechanism that grows the bill in line with value also grows it past what a customer can pay. The providers building these systems instrument the first direction beautifully and the second one barely at all.
Every dollar of spend is a dollar of value captured. The trouble starts the month the customer cannot absorb the dollars they spent.
The front door of usage-based pricing is heavily engineered. The back end, the part that catches whoever spent more than they can absorb, is mostly a blank box.
My take
I spend my days on the back end of consumer credit, the side that deals with the person who cannot pay this month. So when someone says treat pricing like a product, I ask the question I always ask: what happens to the people it is hardest for?
In consumer credit, a bill the customer cannot absorb is not an edge case. It is an entire discipline: hardship programs, forbearance, collections, built over decades with regulation, staffing, and hard-won process behind it, because everyone learned the expensive way that the back end is where trust is actually kept or lost. In usage-based software, that same event, a customer who spent more than they can pay, is handled by a budget cap and an alert email.
The arcade story in the talk is told as a cute warning: hand a seven-year-old your credit card and watch the money vanish in minutes. I hear it as the entire job. The kid with the credit card is not a footnote, she is the customer on the bad month, and a serious monetization platform is one that is built around her, not one that tips its hat to her on the way to talking about margin.
And the talk's center of gravity is speed: cannibalize yourself, change prices weekly, kill the committee, give one person the mandate to ship faster. As growth advice it is mostly right. But notice that every incentive in that room points the same way. The chairperson is hired and fired on rollout velocity. Finance is playing the margin game. Product ships daily. Not one of those roles is measured on whether the customer who overshot gets a humane way back. That job does not exist in the org chart the talk describes, and the absence is the tell.
The meter is the front door. The month it runs away from someone is the back end, and it is the part I would staff first.
Usage-based and agentic pricing are not a fad and they are not wrong. The value alignment is real, and the teams building granular, real-time visibility are doing necessary work. The question is the one I put to every financial system: as this scales to agents spending autonomously at machine speed, does the back end scale with the front door, or does the gap just widen? The arcade has no closing time and the card has no limit unless somebody builds one. The platform that wins the next phase will not be the one with the most precise meter. It will be the one that treats whoever overspent, human or agent, as a first-class user owed a real path back, rather than a support ticket and an apology credit. That is a product decision, and at the moment it is nobody's job.
Threads worth pulling: a talk at Stripe Sessions 2025 on treating pricing as a product, Scott Woody (Stripe, formerly Metronome) in conversation with Shaa Alagumuthu (Anthropic); plus the usage-based billing and cost-governance literature around it. Claims here are paraphrased from my notes on that talk and are my own synthesis, read through the consumer-credit back end I work in.