Key takeawaysA SaaS pricing model built on seats worked for two decades because headcount and software usage moved together, and AI agents break that correlation completely.
Four pricing models are competing to replace or supplement seats right now: pure consumption, token billing, bundled credits, and hybrid structures that layer usage on top of a subscription core.
Adoption is already mainstream, not experimental. Metronome and Monetizely put usage-based and hybrid pricing at 85% adoption among SaaS leaders, with 61% already running hybrid models.
Usage-based pricing is not automatically the right answer. OpenView found 42% of companies that avoided it cited difficulty identifying meaningful usage metrics, and Pendo found 38% of companies that tried it reverted to subscriptions.
Hybrid models are winning for a measurable reason. OpenView benchmarking shows hybrid pricing delivers 1.5 times higher net revenue retention than pure subscription or pure usage pricing alone.
The seat stopped measuring what AI actually does
For roughly two decades, enterprise software grew on a simple formula: price per seat, sell more seats as the customer hires more people. The model built Salesforce and ServiceNow into category giants, and it worked because headcount and software usage moved together in a way that made forecasting straightforward for both sides. A sales team of fifty reps needed fifty CRM licenses. A support team of twenty needed twenty help desk seats. The unit of sale and the unit of value were the same thing.
An AI agent breaks that link entirely. It processes insurance claims, drafts contracts, screens resumes, and answers support tickets without a named employee attached to any of that work. A company can run the same volume of customer support through an AI agent with three people supervising it instead of thirty, and the old seat-based proxy for value simply stops measuring anything real. You can charge per seat for the three supervisors, but you have walked away from pricing the actual work the software is doing.
Satya Nadella summed up the shift on a recent Microsoft earnings call, describing a seat as “just entitlement to some consumption.” That framing is now showing up directly in how vendors structure their pricing pages. Bain’s October 2026 research is careful to note that per-seat pricing is not disappearing. Hybrid structures are gaining ground fast, and the companies moving first are not small experiments run by niche AI startups. They are the vendors that set pricing norms for the rest of the market, which means whatever they land on in the next 18 months will shape what every mid-market SaaS company is expected to offer.
What vendors are testing instead of seats
Bessemer Venture Partners tracks this shift closely in its AI pricing framework, and the pattern across AI-native companies is consistent: revenue increasingly ties to API calls, tokens, or compute cycles rather than logins. Four structures are showing up repeatedly across the market, each solving a slightly different problem.
| Model | How it works | Best fit | Example companies |
|---|---|---|---|
| Pure consumption | Price tracks a direct unit of infrastructure or output, metered continuously | Infrastructure and platform products with highly variable workloads | Snowflake, AWS, Twilio |
| Token billing | Separate input and output rates charged per token processed | AI-native products where compute cost scales directly with model usage | OpenAI, Anthropic |
| Bundled credits | A seat-based core plus a prepaid pool of AI capacity layered on top | Established SaaS products adding AI features to an existing subscription | Microsoft, Notion, Atlassian |
| Outcome-based | Price tied to a completed result, such as a resolved ticket or closed task | Products where the vendor can measure and stand behind a specific outcome | HubSpot (Breeze AI agents) |
The economics explain why vendors are experimenting rather than converting outright. AI inference carries real infrastructure cost that scales with usage, so a flat per-seat fee increasingly undercharges heavy users and overcharges light ones, a mismatch that gets worse as AI features become more central to the product rather than a side feature. Usage-based pricing fixes that mismatch, but it also means revenue can swing with customer demand instead of sitting on a predictable subscription curve, which is exactly the kind of volatility that public markets have historically paid a premium to avoid.
The adoption numbers behind the shift
This is not a niche trend confined to AI-native startups. Metronome and Monetizely put usage-based and hybrid adoption at 85% among SaaS leaders, with 61% of companies already running hybrid pricing by 2025. Maxio’s data adds a growth angle worth sitting with: high-growth SaaS companies, those above 40% year-over-year growth, show a 21% median growth rate specifically when running hybrid models, a meaningfully higher number than peers still running pure seat-based pricing. Metronome’s own 2025 report found that 77% of the largest software companies now use consumption pricing as a lever to grow revenue from existing accounts, not just to price new deals, which tells you this shift is as much about expansion revenue as it is about new logo pricing.
Put together, those numbers describe an industry mid-transition rather than one still debating whether to move. The open question for most SaaS leaders is not whether to adopt usage-based elements. It is how to do it without losing the forecasting discipline that seat-based pricing made easy, and that question is where most of the real risk in this transition actually sits.
Why usage-based pricing is not automatically the right answer
The adoption numbers tell half the story. The other half is a long list of companies that tried usage-based pricing and walked it back, and the reasons why are specific enough to double as a checklist for anyone considering the move.
OpenView’s 2022 survey of companies that deliberately avoided usage-based pricing found 42% cited difficulty identifying a usage metric that actually reflects customer value as their primary reason. An AI recruitment tool is the clearest example: its value is finding the right candidate, not the number of resumes it scans, so billing by volume processed bills for activity rather than outcome. Pendo’s 2023 State of Product Report found 38% of SaaS companies that tried usage-based pricing reverted to subscriptions, largely because their own cost structure, support staff, infrastructure, maintenance, did not scale with usage the way their pricing did, leaving them financially exposed in the gap.
The buyer side resists just as hard. A Forrester study found 67% of enterprise IT decision-makers rank predictable pricing among their top three vendor-selection considerations, which is a direct headwind against any pricing model that makes next quarter’s software bill a variable rather than a known number. ProfitWell’s research found 62% of users reduced the time they spent on a platform specifically because of unpredictable-charge concerns, a pattern the industry now calls bill shock. KBCM’s SaaS Survey found usage-priced companies see sales cycles run about 30% longer than peers with simpler pricing, because procurement teams that cannot forecast a cost tend to stall rather than approve it. And Gartner’s research found 73% of B2B buyers prefer a straightforward buying process over innovative pricing features when comparing otherwise similar products, a preference that cuts directly against novel usage-based structures no matter how fair they are in theory.
None of this means usage-based pricing is wrong. It means it is situational, and the companies reverting to subscriptions are usually the ones that picked a usage metric disconnected from value, or underestimated how much enterprise buyers need predictability to get a deal through procurement.
The hybrid middle ground is winning for a measurable reason
This is why hybrid pricing, not pure usage-based pricing, is the model actually pulling ahead in the adoption data. OpenView’s benchmarking found hybrid models deliver 1.5 times higher net revenue retention than either pure subscription or pure usage-based pricing on its own. A subscription core gives both sides the predictability that 67% of enterprise buyers say they need. A usage or credit layer on top captures the expansion revenue that flat pricing structurally cannot, since a customer getting more value from the product has no built-in mechanism to pay more under a flat fee.
The real constraint behind getting this right has very little to do with which model a company chooses on paper. It has to do with whether anyone inside the company can actually see usage clearly enough to price it, forecast it, and explain a bill to a confused customer before that customer churns over it. We think of this as the consumption visibility gap: the distance between what a pricing page promises and what the finance and revenue teams can actually see happening underneath it. Most companies that walk back usage-based pricing are not failing because the model was wrong. They are failing because nobody built the visibility layer that makes the model work.
A short framework for deciding how to price AI features
Five questions cut through most of the noise in this decision, distilled from the adoption data and the reversion data above.
Does the usage metric track what the customer actually values, not just what is easy to measure. If the answer requires explaining the metric before a customer understands their own bill, it is the wrong metric.
Does value arrive steadily or in irregular bursts. Steady, predictable usage favors a subscription. Irregular, high-value moments favor outcome-based or credit pricing that does not punish quiet months.
Do your own costs scale with usage, or are they largely fixed. A pure usage model on top of fixed infrastructure and support costs is the fastest way to destabilize your own margins.
Can you build or buy the metering and visibility infrastructure this requires. Zuora’s Subscription Economy Index found companies running usage-based pricing spend about 30% more on billing systems than subscription-only peers, a cost that needs to be priced into the decision from day one.
Does your buyer’s procurement process tolerate variability. Enterprise sales cycles with strict budget approval processes are the worst fit for pure usage pricing and the best fit for a hybrid structure with a predictable floor.
What this means for buyers evaluating AI vendors
The same checklist works in reverse for anyone evaluating a vendor’s AI pricing rather than setting their own. Ask whether the usage metric is something your team can monitor on your own, not something you only discover at invoice time. Ask whether there is a cap, a pooled allotment, or an overage alert before you hit a surprise charge. Ask what happens during a usage spike, a busy quarter, a product launch, since that is precisely when a consumption-based bill is most likely to blow past what finance approved. A vendor that cannot answer these clearly has probably not built the visibility layer either, and that gap will eventually become your problem, not just theirs.
We work with SaaS companies navigating exactly this transition, building the revenue analytics layer that turns raw consumption data into forecasts finance teams can actually plan against, rather than a pricing page that looks clean until the first confusing invoice lands.
The SaaS era was built on selling licenses to people. The AI era is beginning to price around what the software actually does, and the companies that close the consumption visibility gap first will be the ones setting the pricing norms for everyone else in 2027.
Frequently asked questions
What is usage-based pricing in SaaS
Usage-based pricing charges customers according to how much of a product they actually consume, such as API calls, tokens processed, or transactions completed, rather than a flat fee per user seat. It ties revenue directly to the volume of work the software performs.
Is usage-based pricing replacing seat-based pricing in SaaS
Not entirely. Research from Bain and Metronome shows seat-based pricing is not disappearing, but hybrid models that combine a subscription core with a usage or credit layer are growing fastest, reaching 61% adoption among SaaS companies by 2025.
What is hybrid SaaS pricing
Hybrid pricing combines a predictable subscription fee with a variable component tied to usage, credits, or outcomes. OpenView benchmarking found hybrid models deliver 1.5 times higher net revenue retention than pure subscription or pure usage-based pricing alone.
When should a SaaS company avoid usage-based pricing
Usage-based pricing tends to fail when the usage metric does not reflect customer value, when a company’s own costs are largely fixed rather than variable, or when enterprise buyers in the target market require predictable budget approval. Pendo’s research found 38% of companies that tried usage-based pricing reverted to subscriptions for exactly these reasons.
Learn more:
Revenue Analytics for Revenue Teams
AI Studio for B2B SaaS