From Seats to Credits: The Pricing Revolution
Credit-based pricing grew 126% year-over-year. When Salesforce lost $285B in market cap in 48 hours on the AI seat-replacement thesis, the entire industry got the message. Here is how pricing is being rebuilt.
The Seat Problem
For two decades, per-seat pricing was the default model in SaaS. It was simple to explain, simple to forecast, and simple to sell. Then AI showed up and broke the contract. If one AI agent can do the work of five humans, why would a customer pay for five seats?
Salesforce learned this the hard way. In February 2024, the company lost $285 billion in market cap in just 48 hours after analysts modeled what happens when AI agents replace human users inside CRM workflows. The thesis was straightforward: if Agentforce automates 40% of sales tasks, customers need 40% fewer seats. Revenue collapses, even if the product gets better.
That event sent shockwaves through every SaaS boardroom. The seat-based model has a fundamental alignment problem: it charges for headcount, not for value delivered. When automation reduces headcount, the vendor is punished for making the customer more efficient.
The value metric is shifting from "how many people use it" to "how much value was created." Seat-based pricing penalizes vendors when their product eliminates manual work. The best SaaS companies are rebuilding pricing around outcomes, not occupancy.
The data confirms the shift is accelerating. According to Kyle Poyar at OpenView Partners, credit-based pricing models grew 126% year-over-year across their portfolio. Meanwhile, 61% of SaaS companies now use some form of hybrid pricing that combines a base fee with usage-based components.
Credit-Based Pricing: The 126% Surge
Credit-based pricing is the fastest-growing model in SaaS. Customers buy a bundle of credits upfront, then spend them across features at different rates. A simple API call might cost 1 credit. A complex AI workflow might cost 50. The model decouples price from users and ties it to consumption.
HubSpot is a textbook example of the transition. In 2025, they introduced credits for their AI features while keeping seat-based pricing for the core CRM. A customer pays $50/seat/month for Sales Hub, plus a credit bundle for AI-powered email generation, lead scoring, and content creation. The seat price stays flat. The credit spend grows as adoption deepens.
This hybrid approach lets HubSpot protect their existing revenue base while creating a new expansion vector. The data suggests it is working: customers on credit-based AI features show 23% higher net revenue retention than those on flat plans.
Usage-Based Models: The Churn Killer
Usage-based pricing is not new. Twilio has charged per API call since 2008. What changed is the scale of adoption. Paddle and ProfitWell data shows that usage-based pricing reduces monthly churn to 2.1%, compared to 3.9% for flat-rate models. That is a 46% reduction in churn, which compounds into dramatically different LTV curves.
The mechanism is straightforward: when price scales with usage, customers who are getting value automatically pay more. Customers who are not getting value pay less, but they do not churn entirely. They stay on the platform at a lower spend, available for reactivation. Flat-rate models force a binary choice: pay full price or cancel.
| Model | How It Works | Pros | Cons | Example |
|---|---|---|---|---|
| Seat-Based | Fixed price per user per month | Predictable revenue, easy to sell | AI reduces seat count, misaligned with value | Salesforce CRM |
| Usage-Based | Pay for what you consume (API calls, storage, compute) | 46% less churn, scales with value, low entry barrier | Revenue volatility, harder to forecast | Twilio, Snowflake |
| Credit-Based | Buy credit bundles, spend across features | Predictable + flexible, 126% YoY growth | Complex to design, credit expiry friction | HubSpot AI, Zapier |
| Hybrid | Base platform fee + usage or credits | Revenue floor + expansion upside, 61% adoption | More complex pricing pages, harder to quote | Datadog, MongoDB |
Snowflake is the poster child. Their consumption-based model means customers pay for compute credits and storage used, not for seats or licenses. Revenue is volatile quarter to quarter, but their net revenue retention consistently exceeds 130%. When customers grow, Snowflake grows automatically.
Datadog runs a similar model for infrastructure monitoring. Customers pay per host, per million log events, and per million traces. The beauty is that as a company scales its infrastructure, Datadog usage scales in lockstep. There is no renewal negotiation - the meter just runs.
The contraction risk is real, though. MongoDB saw revenue decelerate in Q3 2024 when several large customers optimized their queries and reduced storage consumption. Usage-based pricing means revenue drops when customers become more efficient. The best companies mitigate this by continuously shipping new features that generate new usage vectors.
AI Pricing: Four Models Emerging
AI features have introduced entirely new pricing questions. The cost structure is different (inference costs are real and variable), the value delivery is different (outcomes vs. access), and customer expectations are different (they want to pay for results, not compute).
Four distinct AI pricing models have emerged, each suited to different product types and customer segments.
Charge based on input/output tokens consumed. Used by API-first products.
Charge when the AI delivers a measurable result (lead scored, ticket resolved, document processed).
Charge per workflow execution, regardless of token count. Bundles complexity into a flat per-run fee.
Pre-purchased credits that can be spent across multiple AI features at varying rates.
Per-outcome pricing is the model the research shows will win long-term. It perfectly aligns vendor and customer incentives: the vendor only gets paid when the AI delivers measurable value. Intercom's Fin charges $0.99 per resolved support ticket. If Fin cannot resolve the ticket, Intercom earns nothing. That is the kind of confidence that collapses buying objections.
NDR Impact: Why Pricing Model Determines Growth
Net Dollar Retention is the single most important metric in SaaS, and pricing model is its biggest lever. Usage-based and credit-based companies consistently post higher NDR because revenue scales with customer growth automatically. There is no upsell conversation required - the customer simply uses more.
Snowflake has maintained NDR above 130% for years. Datadog regularly posts above 125%. Compare that to pure seat-based companies where NDR expansion requires sales-assisted seat additions, typically landing between 105-115%.
| Pricing Model | Typical NDR | Expansion Mechanism | Example |
|---|---|---|---|
| Seat-Based | 105-115% | Sales-assisted seat additions | Salesforce |
| Usage-Based | 120-140% | Automatic with customer growth | Snowflake (130%+) |
| Credit-Based | 115-130% | Increased credit consumption | HubSpot AI |
| Hybrid | 115-135% | Base + organic usage growth | Datadog (125%+) |
The math is unforgiving. A company with 130% NDR doubles its revenue from existing customers in roughly 2.5 years, even with zero new customer acquisition. A company with 110% NDR takes over 7 years to achieve the same result. Pricing model is not a finance decision - it is a growth strategy decision.
Pricing Frameworks: How to Choose
There is no universal best pricing model. The right choice depends on your product, your customer, and your cost structure. The research suggests three questions that determine the answer.
MediaSeize Analysis
The pricing reset is not optional. Every SaaS company will face the same question Salesforce faced: what happens to your revenue when AI reduces the number of humans using your product? The companies that answer this question proactively - by decoupling price from seats - will thrive. The ones that wait will find themselves explaining revenue contraction to their board.
We recommend a phased approach. Do not rip out seat-based pricing overnight. Instead, layer a credit or usage component alongside it. HubSpot did this well: the CRM stays seat-based, but AI features run on credits. This protects existing revenue while creating a new expansion vector that is aligned with how AI actually delivers value.
The 61% hybrid adoption number tells the story. The market is not choosing between seat-based and usage-based. It is combining them. The base fee provides revenue predictability. The usage component provides expansion upside. Together, they create a model that works for both the CFO (who wants forecasting accuracy) and the CRO (who wants NDR above 120%).
Three moves to make now:
- Identify your value metric. If you charge per seat, ask: what does a seat actually produce? Leads generated? Tickets resolved? Reports created? That output is your new pricing anchor.
- Model the contraction scenario. What happens to your revenue if AI cuts seat count by 30%? If the answer is "revenue drops 30%," you need a new model before your customers figure that out.
- Ship a credit-based AI tier. Even if the rest of your pricing stays flat, give customers a way to buy AI capabilities on a credit basis. It is the lowest-risk way to test usage-based expansion.
The Salesforce $285B wakeup call was not a one-time event. It was a preview of what happens to every seat-based company that does not adapt. The pricing revolution is here. The data suggests moving early, not waiting for the market to force the change.
Ready to put these frameworks to work?
MediaSeize builds the growth systems described in this report for CPG, supplement, and DTC brands. Tell us about your brand and we'll follow up within 24 hours with specific thoughts on where to start.