Mostly metrics is proudly powered by Brex
The CFO role has evolved faster than the tools built to support it.
Most finance teams are still running infrastructure designed for a job that no longer exists. The reporting. The reconciling. The close that bleeds into the next month. That's not finance - that's overhead with a title.
Agentic Finance shouldn't multiply your output - it should eliminate the work that was never worth doing in the first place.
That's why I run Mostly Media on Brex - an intelligent finance platform with AI-powered agents that do exactly that. Expenses handled automatically, policy enforced before the spend happens, books closed in minutes. So I can spend my time on the work that actually moves the business.
Job Stuff
Gee wiz. I run a recruiting company now. Life comes at ya fast.
We do finance hiring. And we place at the manager, director, and VP levels.
If you're hiring, work with us here.
If you're passively open to new roles, get into the ecosystem here.
Signs You Should Move from SaaS to Usage Based Pricing
👋 Hi, it's CJ Gustafson and welcome to Mostly Metrics - my newsletter for CFOs about the metrics and financial strategies behind the world's best businesses.
And you know what really grinds my gears?
When people talk about usage-based pricing like it's purely a pricing move. As if you can wake up and announce:
"We charge like Planet Fitness today. I want to charge like a taxi cab tomorrow."
That ain’t how it works.
Going from a subscription to a usage or commitment model is a one-way door.
Perhaps you’ve heard of this concept of one-way and two-way doors before. A two-way door is a decision that’s reversible. Hire the wrong CRO? You can go out and get another one. Sucks, but it’s not existential.
(They usually last 18 months anyways.)
A one-way door is much harder to walk back through. Rewriting the code of your flagship product is incredibly difficult to undo once you’ve started. Changing to usage based pricing is also a move with little recourse.
You're rebuilding how the product gets delivered, how your reps get paid, and how you forecast and recognize revenue. Whole departments end up with new jobs, and new departments are created.
So the interesting question is less whether you'd like to charge based on usage (most founders would happily take the expansion story, given how seat based models are being dragged down with headcount compression). The real question is whether the cost of serving a customer has crept up to the point where you can't afford not to.
And you find that answer in your gross margin…
I've now talked to three CFOs who walked through this exact door, at Confluent, Figma, and Couchbase. And they all started by investigating one place: gross margin.
Confluent saw it coming in their own COGS
Confluent is a great company to study because most of us started out selling the way they did (myself included). Back in 2019 they were monetizing in three ways at once:
a product-led motion off their open source community,
a traditional on-prem sale with the usual bookings and rev rec, and
a cloud-native product where they also had to run the customer's instance.
The first and last ones are where the costs began to pile up based on how customer’s interacted with their product. Think of Confluent as a digital conveyor belt that continuously streams data between different apps and systems. Given the data component, it doesn’t make sense to charge a fixed price based on number of humans.
When I had Confluent's CFO, Rohan Sivaram, on Run the Numbers, he was blunt about why the move was, in his words, a no-brainer:
"We are software execs. How we grew up, we never thought about gross margins, because the incremental cost of shipping a unit of software was zero. But in a cloud business, the incremental unit of consumption had real costs associated with it... If we don't have consumption or usage-based pricing, I don't think we'll be able to have control over the unit economics of the business."
I've lived in that blind spot he's describing. I spent years in FP&A at software companies where the marginal cost of one more login rounded to approx zero. Our COGS were pretty much just Customer Support + Jira Tickets. In a cloud business, the incremental unit of consumption is also an incremental unit of cost. So while there is more upside, you need to play much better defense.
And realizing this trend wasn’t enough. They knew they had to make the change, but were playing on hard mode, as there wasn’t much out there to copy. It’s always easier when you have a number of comps to look to.
"When we went through this transition, none of us had any pattern recognition. It was new. Five years back, there were probably two public companies doing it at scale outside the hyperscalers: Snowflake and MongoDB. Databricks was doing it, but they were private."
In many ways, these companies were making it up as they went. That includes sales comp plans, sliding discounts for commitments, and rev rec. So, like, basically inventing new ways to get paid and paid people.
And now all the AI companies owe them a Christmas card for going first.
AI made usage impossible to ignore
Rohan doesn't think this stayed a Confluent problem. He thinks AI has now turned it into everyone's problem:
"Fast forward five years, and every AI-native business has some form of usage-based or consumption-based pricing. It is needed. Otherwise you're probably going to lose money. You won't have control over how the unit economics eventually shape up."
You can watch that land in real time within Figma's gross margin. When I had their CFO, Praveer Melwani, on the show, he addressed the trade:
"If you looked at the business in Q1 of last year, prior to rolling out our AI features, we were like 90, 91-ish percent gross margin... we exited at about 86% gross margin end of Q4, and that's because we were serving the cost of inference to power some of our new products and features."
So five points of margin as they worked AI into their traditionally seat based product. To be clear - Figma very much expected this, and was actually greasing the skids for longer term user adoption.
"Giving people some opportunity to try our AI features by embedding a set of AI credits across all of our seat types was extremely important to us, because you're not going to experience the magic unless you try. But if you're using the product in more mature ways, we want to give ourselves an opportunity to monetize some of that."
That’s why they took it a step further to package their usage based pricing as “AI credits”. this aligned everyone to get the most value out of the product.
"The cost to serve these AI products is non-trivial. If you don't have the ability to align your model to one where it makes sense to drive more utilization on platform, and that's actually good for the business, you're going to end up making these weird local decisions. Like, ‘do I actually want people to use my AI products or not?’ That doesn't feel right."
Said another way, under flat pricing, every time a customer leans on your most expensive AI feature it costs you money, so part of your brain starts rooting against your own product getting used. That's a trap Figma wanted to avoid. So they came straight from their product-led roots: put credits in everyone's hands, then monetize the heavy users.
I asked him about the longer term plan, and he wasn't sweating the exact margin landing spot either, because a consumption model self-corrects. It's not a perfect "tokens in as dollars out," as he put it, but it's heading that way, with a credit drawdown that flexes with whatever each request actually costs.
Why it's a one-way door
I asked Rohan straight up whether he saw it as a one-way or a two-way door:
"It is a one-way door. I don't think it was reversible. And it's not just the pricing... it's your whole org, how you're architecting the org.
[Given the importance of the decision]
The trifecta was the president of field operations or CRO, your chief product officer thinking about product pricing, and the CFO, just joint at the hip."
Rome’s Usage Based Pricing Triumvirate
Greg Henry made the same move as CFO at Couchbase, taking a historically enterprise-subscription business into a managed-service consumption model. I asked him what changed beyond pricing:
"We thought of it right away as two different business models. We had to build operations around a consumption business: tracking consumption, who's over-consuming, who's under-consuming, why. Sales compensation plans, the way we track in finance. Everything changed."
"Feel the rhythm, feel the rhyme, get on up, it's bobsled time!”
The rhythm of the job changes, too. In a subscription world, the docusign sig is the climactic end to the story: you know the TCV, you know what you're recognizing, the rep gets the commission, kick it over the fence to customer success for a firm calendar based renewal down the road. In a consumption world, the signature is merely the kickoff. Here's Rohan on what that does to the calendar:
"In a consumption business, once you have bookings, fantastic, you have a customer commitment. But you need to forecast every customer's individual consumption curve. So with my morning coffee, it's always looking at what yesterday's consumption was, tracking it, looking at outliers. And that's not just me. It's the broader business leaders too."
Greg put the same point from the sales side. There's no renewal date to circle on the calendar anymore:
"There's little chance they're going to come back exactly 12 months later and buy more. They're probably going to come back sooner, because they're consuming."
Where to actually look
So here's where I'd point you. Pull up your own financials and watch three things, in roughly this order:
Your variable cost to serve a customer across all your sales motions
The cost to serve your smallest customers vs your largest ones
Your gross margin movements (net any employee changes in CS) quarter to quarter
When all three of these are degrading, the decision has mostly been made for you. The only thing left to choose is whether you walk through the door on your own terms, or get shoved through it on someone else's.
Run the Numbers Podcast
On this episode of Run the Numbers, I sit down with Confluent CFO Rohan Sivaram to talk goal setting, prioritization, consumption-based pricing, hybrid zero-based budgeting, and the frameworks finance leaders use to scale companies.
Rohan shares why he carries his 12-month goals with him, how he evaluates opportunities through TAM, technology, and team, and why usage-based pricing changes the entire operating model.
Quote I’ve Been Pondering
“What’s the point in being an outlaw when you got responsibilities?”
Wishing you gross margins that move for reasons you picked,
CJ
