Usage-based pricing is the default recommendation for AI products, and the default is wrong.
Cursor switched its Pro plan to token-based billing in June 2025 and apologized within three weeks. Users who budgeted $20/month saw invoices hit $60 to $100, and one five-person team spent $4,600 in six weeks. Snowflake customers routinely report hidden consumption growth from cross-region data sharing, spending far more than they forecasted. Meanwhile Notion AI launched at $10/member/month as a flat add-on, hit 65% workspace adoption within a year, and reduced churn enough to save an estimated $30M annually.
Two of those pricing structures created billing crises. The flat one kept its customers.
Five Variable Line Items at Once
Three years ago, a typical company's software budget had 14 fixed line items and one variable cloud bill, and the CFO could forecast Q3 spend in an afternoon. Today, half those tools have shifted to consumption pricing.
| Budget Line | 2023 (Fixed) | 2026 (Variable) |
|---|---|---|
| Code editor | $20/seat/mo | Per completion (Cursor) |
| Support tool | $50/seat/mo | $0.99/resolution (Intercom Fin) |
| Analytics | $500/mo flat | Per query |
| Data warehouse | $2,000/mo flat | Per compute credit (Snowflake) |
| AI assistant | $30/seat/mo | Per message |
One variable line item is manageable. Five variable line items compounding against each other is unplannable, and that's the budget reality your pricing model lands in.
78% of IT leaders report unexpected charges from consumption-based AI pricing, and 90% of CIOs cite cost forecasting as their top challenge in AI deployment. These aren't complaints about any single vendor. They're complaints about what happens when an entire software stack becomes variable at the same time.
The company that offers predictability earns a structural advantage at renewal. Not because the product is better, but because the invoice is forecastable.
Two Ways Renewals Break
Matt Green, who studies pricing across 500+ SaaS companies at Growth Unhinged, captures the core tension: usage-based pricing is easiest to close and hardest to renew.
The sales motion is genuinely simple. "Only pay for what you use" removes objections, carries no shelfware risk, and asks for low commitment. Easy first signature.
The renewal is where it breaks, and it breaks in two opposite directions.
Bill shock is the failure mode where the product worked too well. The customer adopted enthusiastically, usage spiked, and the invoice arrived at 3x expectations. This is the Cursor pattern: users didn't blame themselves for using the product, they blamed Cursor for the surprise.
Perceived waste is the failure mode where the product didn't get adopted. The team used it lightly and the invoice was low, which sounds fine until budget review, when the low invoice tells the CFO the product isn't essential and it becomes the first line item cut. Low usage signals low value, even when the product delivered exactly what was needed.
Green reports an executive revenue leader whose gross revenue retention actually improved after switching from consumption back to per-seat pricing. That's not a theoretical argument. It's an observed result, and it's the single most inconvenient data point for anyone insisting seats are dead.
Usage Pricing Punishes Your Best Customers
The customer who uses the product most gets the highest bill. The team that drives adoption across every department gets the largest invoice. The power user who builds workflows around your tool pays more than the person who logs in once a month.
That's the opposite of a loyalty reward, and enterprise buyers especially hate it because they can't forecast costs for a product they plan to use more over time. A successful deployment means a bigger bill, a growing team means a growing invoice, and the incentive structure quietly pushes customers to limit adoption rather than expand it.
The companies moving from pure usage to hybrid aren't retreating from value-based pricing. They're responding to what customers actually want, which is predictability with value alignment. The customer wants to know what they'll pay, and they want to feel that what they pay reflects what they get. Hybrid pricing gives them both. Pure usage gives them neither.
Usage Pricing Frames You as a Commodity
Vin Vashishta makes the point most pricing discussions skip entirely. When you price per token, per API call, or per credit, you frame your product as infrastructure, and infrastructure gets compared on price per unit. The customer shops for cheaper tokens the way they shop for cheaper cloud compute, and the moment a cheaper alternative appears, the switching calculus is simple arithmetic.
An outcome gets compared on results. The customer evaluates whether the problem got solved, not how many tokens it consumed, and switching means retraining, reintegrating, and accepting the risk of worse results.
| Pricing Frame | Customer Evaluates | Switching Trigger |
|---|---|---|
| Per token / per credit | Cost per unit | Cheaper alternative appears |
| Per resolution / per outcome | Problem solved or not | Results decline |
| Plan with guardrails | Total value vs. total cost | Budget review, competitive feature gap |
Commoditization accelerates this. Model costs drop roughly 10x every 18 months, and when you price per token, every cost reduction puts direct downward pressure on your revenue. Plan-based pricing decouples your revenue from the model cost curve: when inference gets cheaper your margins improve, the customer sees the same bill, and you capture the efficiency gain instead of passing it through.
Where the Market Actually Landed
Hybrid pricing adoption surged from 27% to 41% in 2025 and is projected to reach 61% by end of 2026, while pure seat-based pricing dropped from 21% to 15% over the same period. The market isn't choosing between seats and usage. It's choosing both.
HubSpot moved its Breeze AI agents from $1.00 per conversation to $0.50 per resolved conversation in April 2026, shifting from activity to outcomes. Intercom charges $0.99 per resolution but bundles it with per-seat helpdesk pricing, a predictable base with a variable outcome layer on top. Notion bundled AI into its Business tier entirely in May 2025, absorbing AI cost into a single forecastable line item.
Each moved toward more predictability, not less.
What Actually Works: Plans With Usage Guardrails
The companies reporting the best retention numbers share a structure.
| Component | Purpose |
|---|---|
| Base plan ($X/month) | Predictable cost the CFO can budget |
| Included usage allowance | Covers 80-90% of customers without overage |
| Published overage rate | Clear per-unit cost above the allowance |
| Hard spending cap | Maximum monthly bill, no surprises |
Notion AI at $10/user/month with included responses achieved 90% renewal rates and 91% monthly active retention. Jasper and Fireflies.ai use flat-rate add-ons or seat-based AI plans to avoid billing anxiety. Anthropic offers team seats from $25/seat/month with usage limits per tier, keeping API-rate billing separate from the workspace product.
It's worth being precise about what Cursor's backlash was actually about. The anger came from removing the plan, not from adding usage. Users accepted usage limits; they rejected unpredictable bills. The plan creates the anchor, the guardrails handle edge cases, and the cap prevents bill shock.
The Cost Visibility Gap Underneath
Every argument about seats versus usage versus outcomes traces back to the same gap: most companies don't know their per-customer cost-to-serve.
Without that number, the pricing model debate stays abstract. You can't set a confident flat price without knowing your cost floor, can't set a confident usage rate without knowing your cost variance across customers, and can't price on outcomes without knowing what outcomes cost to produce.
OpenAI, Anthropic, and Vercel all surface granular usage dashboards for teams to model ROI, because cost visibility is the prerequisite to any pricing structure. The 78% of IT leaders reporting unexpected AI charges aren't facing a pricing model problem so much as a visibility problem, and the same is true on the vendor side: every model swap and prompt change shifts the cost underneath the price, whether or not anyone measures it.
Some customer segments are wildly profitable under flat pricing. Others need guardrails. A few might justify outcome-based pricing where the outcome is measurable and consistent. You can't know which until you see the numbers customer by customer.
Pricing Determines Behavior, Not Just Revenue
The standard framing treats pricing as a revenue capture mechanism: pick the model that extracts the most value. That framing is incomplete, because pricing also teaches your customers how to behave. Usage-based pricing teaches them to limit consumption. Plan-based pricing teaches them to maximize adoption within their tier. Outcome-based pricing teaches them to demand measurable results.
The Copilot story showed what happens when a company can't see its own cost distribution, and the Cursor story showed what happens when customers can't see theirs. Both problems share a root: insufficient visibility into what each customer costs and what each customer gets. The model you choose isn't just a billing decision. It's a product decision that shapes how customers use what you built.
Visibility into cost-to-serve is what lets you design the right hybrid instead of guessing at one. Bear Lumen gives you that per-customer view automatically, so the allowances, overage rates, and caps in your pricing come from your own cost distribution rather than from a competitor's pricing page.