Growth Unhinged is proudly supported by Metronome, a Stripe product
Everyone wants to talk about outcome-based pricing for AI (myself included). Metronome CEO Scott Woody thinks it's mostly a myth. In a new post, he explains why only companies with monopoly-like power or very large contracts can measure outcomes well enough to price on them.
For everyone else, Scott argues for pricing on outputs: objective, countable units of work like a generated image or a resolved support conversation. He also makes the case for why token billing works better behind the scenes than on your invoice, and how unified credits bridge the gap. Read Scott's full post.
👋 Hi, it’s Kyle and welcome to Growth Unhinged, my weekly newsletter exploring the hidden playbooks behind the fastest-growing startups.
It’s hard to believe we’re only a week away from the second State of B2B GTM report launch, the definitive benchmark for what’s working in GTM. Maja Voje and I are revealing the findings live on October 14th and we’ll be joined by speakers from Lovable, Granola, Vercel, and more. RSVP to join us.
It’s been nearly four years since ChatGPT launched. Since then AI products have fully taken hold of our working lives (and our personal ones).
Yet monetizing AI is still not settled. I’ll show you the receipts. In just the past couple months:
OpenAI began testing outcome-based pricing with select enterprise customers, as The Information reported.
Cognition, the AI engineering startup newly valued at $48 billion, introduced an AI productivity guarantee where they’ll fund usage of up to $10 million until AI delivers real engineering value (the methodology behind this is pretty interesting).
Automation startup Gumloop rolled out what they’re calling “a simple, transparent pricing model” built on a pass-through model: they pass through tokens and compute at cost and add an 8% orchestration fee. (Clay went in a similar direction in March, as I covered earlier.)
Salesforce’s deputy CFO told an investor conference that their pricing experiments are “an anxiety-filled architecture right now of different pricing structures and contract frameworks.” (Salesforce is increasingly talking about outcome-based pricing, but also offers license models, credit models, and fixed price ELAs.)
So where are we headed: outcome-based pricing, pass-through pricing, choose-your-own-adventure pricing? Yes.
The overarching theme is from charging for access to charging for usage and outcomes. Data from my 2026 State of Monetization report found that 76% of AI-native companies already have a form of variable or hybrid pricing, compared to 49% of SaaS companies.
Investors used to hate variable and usage-based pricing. Now they're pushing for it.
When asked which pricing model would be preferred by investors, only 15% said flat-fee or seat-based subscriptions. The remainder said investors prefer variable pricing models: usage-based (24%), outcome-based (26%), or hybrid (35%).
Beware that shifting pricing models can be a hard pivot. It impacts billing, sales comp, enablement, product strategy, post-sales investment, forecasting, and more. Today I’ll unpack conversations I’ve been having with founders and revenue leaders about what needs to change when adopting usage-based pricing.
Theme 1: Customers still want predictability even with usage-based pricing
When people hear the term usage-based pricing, they immediately think pay-as-you-go. That’s not what (most) customers want to buy.
The message of “you only pay for what you use” sounds compelling right until the first overage bill hits.
The GTM focus needs to be about making variable pricing feel predictable. This work has the added benefit of making revenue more predictable, too.
There are two main routes to address this:
1. Adopt a universal credit model
HubSpot has launched new AI capabilities that span sales, data enrichment, support, content generation, and even payment collection.
Each of these has its own pricing metric. Support is priced based on successful AI resolutions. The prospecting agent is priced per lead. You could imagine how hard it would be for a buyer to estimate exactly how much they’ll need to spend.
HubSpot uses a universal credit mechanism to take away some of this complexity. Each action gets converted into credits. Customers can buy a pack of credits, then draw these down flexibly across products as they choose. Think of this like a retainer with a law firm: you don’t know exactly how many billable hours you’ll need, but you can estimate the overall t-shirt size and then true-up later.
Once a universal credit mechanism is in place, just about anything could be folded in. Perhaps in the future a “seat” might even get subsumed under the credit model, as Steven Forth suggested.
2. Make your pricing structure more predictable
The structure you put around your pricing can go a long way toward making buyers more comfortable. Four structures to consider are:
Hard budget caps. This gives buyers peace of mind by capping their potential usage/spend, ideally at both the account and user-level. (For inspiration, see AWS’s revamped getting started experience.)
Usage-based tiers with an annual drawdown: Customers commit to a certain level of usage upfront; however, this usage can be consumed flexibly over the year, like a gift card. This gives customers plenty of time to see what steady state usage looks like before approaching their limit or an overage bill.
Platform fee (includes usage) plus usage. Also known as a three-part tariff, this model has a subscription fee with a baseline of usage included “for free”. Providing a minimum amount of usage helps get the customer hooked and usually incentivizes more overall consumption.
Adaptive flat rate. The customer commits to a usage-based tier, but can use the product as much as they want with no overages or upgrades during that contract. True-up happens the following year based on actual usage. (The downside is that you’re on the hook for any extra usage costs!)
Theme 2: Customers aren’t refilling the tank fast enough
Usage-based companies have less room for shelfware compared to a traditional subscription businesses. If you oversell the customer, they can see it right away. The better path from an LTV perspective is to start with a high value use case, nail it, and then get the customer comfortable rolling out wall-to-wall.
Think of this like your advertising spend. You probably won’t put $1 million into Meta or ChatGPT ads in your first month. But you’ll put in $5k, measure ROI, then go big once there’s proof.
During HubSpot’s Q2 2026 earnings call, CEO Yamini Rangan admitted that “the quarter we expected did not fully materialize” — and deliberate pricing changes were a reason why. HubSpot introduced agent trials, which extended the buying process, but gave customers more confidence in their AI ROI. “Customers adopting AI want proof of value before they commit and predictability in what it costs,” Yamini said.
There are two focus areas for getting customers to refill the tank faster:
1. Time-to-ramp
Adoption needs to happen before expansion can materialize. But the customer isn’t always motivated to move quickly on the initial deployment.
Time-to-ramp helps you monitor how long that takes the average customer to reach 80% of their usage allowance. Measure it, assign an owner, and then run experiments to accelerate it.
2. Share of wallet
Expansion has a ceiling within each account. Measure how much spend is addressable and how much you have (share of wallet). Then prioritize expansion efforts accordingly.
Fin, the AI agent for customer service, charges $0.99 per outcome. They only make money when the product works. Co-founder Des Traynor told me he closely monitors Fin’s total automation rate. This quantifies how much work Fin completes relative to the total addressable work in an account. “Customers might not expose Fin to certain topics, cases, or support channels like voice or WhatsApp. We measure how much of the support volume we’re getting and what percentage of that work we’re resolving.”
Theme 3: Keep reps motivated without overselling (or overpaying)
Sales compensation is a perpetually tricky topic for usage-based businesses.
The default comp approach is to pay based on new bookings. This starts to break down as either (a) reps oversell on the initial deal in order to maximize bookings or (b) reps feel cheated for not sharing in the upside as customers expand.
Mature cloud companies shifted from bookings to actual consumption. Snowflake, for example, made a big (and public) sales comp change from bookings to a hybrid of bookings plus consumption.
In theory this better aligns the rep’s incentives with the target business outcomes. In practice, it presents its own challenges:
Reps miss out on ‘ring the gong’ moments since consumption is gradual rather than a big one-time event.
Reps are incentivized to chase expansion revenue rather than net-new deals, which take more effort for less initial quota impact.
Reps get paid even when they don’t do anything to drive expansion.
Three paths to keep reps motivated without overselling:
1. Introduce a new logo component to commissions
Snowflake itself did this in FY2025, when it broke out rep compensation into bookings, consumption, and a new logo component. Net-new customer adds went up by 36% after the change, following a flat FY24 to FY25.
What this could look like: a fixed $ payout per new logo regardless of initial deal size. (I recommend adjusting the payout for SMB, midmarket, enterprise, and strategic accounts.)
2. Pay based on estimated revenue
Another approach is more operationally intensive, but also more precise. Some companies will compensate sales reps based on the estimated revenue, which comes from a combination of committed bookings and expected pay-as-you-go or overage revenue. This is usually paired with true-ups as actuals come in above or below the estimated amount.
The estimated revenue approach allows sales to share in the upside on their deals while still encouraging upfront committed bookings.
3. Distinguish between organic and inorganic expansion revenue
Not all expansion revenue is created equal. Consider delineating what’s “organic” versus “inorganic” and design separate motions to go after both opportunities:
Organic expansion: Your customers simply use more of your product(s). This is a by-product of customers getting onboarded, seeing value from your product, and wanting to use more of it. The organic expansion number is usually owned by customer success rather than sales. That’s why many companies avoid overpaying commission on this expansion after the customer’s first year.
Inorganic expansion: Your customer bought new products or added use cases / business units that were not part of the initial purchase. Generating inorganic expansion generally requires a commercially-minded rep to build an account plan, spot untapped opportunities, ask for intros to other teams, and co-create business cases. Treat these opportunities more like you would a new business sale.
Thanks for reading Growth Unhinged! To support my work and access the full archives of 200+ posts, please consider upgrading to a paid subscription.
What to check before you go usage-based
Usage-based pricing is (very) trendy right now. That’s not a reason to adopt it.
Before you touch your pricing, answer these questions:
Can my customers predict their bill? If not, consider a credit pool, an annual drawdown, or an adaptive flat rate first.
Do I know how fast customers ramp and how much of their wallet I’m capturing? If not, start measuring these and assign owners for improving them.
Will my comp plan reward the behavior I actually want? If reps still get paid 100% on bookings, expect oversold deals and stalled expansion.
Don’t change your pricing because of your investors. Change it because of your customers.
🔓 Bonus for paid subscribers:
Download the full 37-page State of B2B Monetization report for real data from 230+ companies on pricing model adoption, trends over the last year, and what’s working in monetization.
Access all pricing resources including Claude skills and project plan templates in the premium Members Library.
Subscribe to Kyle Poyar's Growth Unhinged to read the rest.
Become a paying subscriber of Growth Unhinged to get access to this post and other subscriber-only content.
UpgradeA paid subscription gets you:
- Full archive
- Subscriber-only bonus posts
- Subscriber-only discounts and perks
- Full Growth Unhinged resources library