Should you move to usage-based pricing? A decision tree for legacy B2B software
The best reason to move to usage-based pricing is that your customers want their bill to grow when their business grows. A fixed fee tied to capacity or volume charges them for headroom they may never use, and it charges your biggest accounts too little. If customers have been asking for that kind of alignment for years, you have a strong case.
Most B2B vendors are feeling a different kind of pressure, though. AI inference costs are eating into margins, investors want to know why you're still selling seats, and a competitor's new credit plan has your sales team nervous. Each of those calls for a different answer, and only some of them lead to usage-based pricing.
This guide is for people running established licensing and subscription businesses. It walks through what's actually pushing you toward a change, where margin compression fits in, a decision tree for the pricing model itself, and the packaging and operational questions that decide whether the new model makes money once it's live.
The short answer
Move to usage-based pricing when the usage metric tracks the value the customer gets, your cost to serve moves with that same metric, and customers can predict and control what they'll pay. If only the first of those holds, add usage to your subscription rather than replacing it. If none of them hold, keep your current model and change how you package it.
What matters more than the model you pick is how easily you can change it. A vendor that can test a new offer every month without waiting for an engineering release will learn more in a year than one that picks the "right" model once and leaves it alone.
Start with what's driving the change
Before you choose a model, be honest about who is asking for it.
When customers ask, that's the healthiest reason to move, but there's a catch. They ask for usage-based pricing in simple terms, and once they have it they want protection when their usage drops and the upside when it spikes. So your first job is to find out what they meant. Often it's "let me start smaller" or "stop charging me for capacity I don't use," and a subscription with included usage answers that better than pure pay-as-you-go does.
When a competitor repackages around credits or outcomes, buyers can't compare your prices directly anymore. That's the pressure most dangerous to ignore and the easiest to overreact to. If you copy a competitor's model without checking it against your own value metric, you take on their economics without getting their advantage.
When AI features add a variable cost to every action, the pressure is real, but it's a cost-alignment problem before it's a pricing-model problem, as the next section explains.
And when investors push, treat it as context. Kyle Poyar's 2026 State of B2B Monetization report found only 15% of respondents think investors prefer flat-fee or seat-based subscriptions. That tells you where sentiment is heading. It doesn't tell you what your customers need, and they're the ones who should drive a repricing.
Where margin compression comes in, and who it affects
AI changes the cost structure of software. Bessemer Venture Partners' AI pricing playbook puts gross margins for AI companies at 50 to 60%, against 80 to 90% for traditional SaaS. Every model call costs you money, so a flat seat price can lose money on your heaviest users.
When AI costs eat into margin, it's usually because what you charge for and what drives your costs have moved apart. Usage-based pricing can bring them back together, but it often isn't the best tool for it. Start from value, and ask whether your pricing metric matches the value the customer receives, and then whether your cost drivers, such as AI inference, line up with that value.
If your answer to the cost problem is to mark up processing, you've moved into a low-margin business. A vendor that passes tokens through with a margin on top has become a reseller of someone else's compute, while pricing tied to what the software does for the customer keeps software margins.
Margin compression hurts whoever pays for the compute. That makes it most acute for:
- •
Vendor-hosted SaaS and AI products, where every query, agent run or data job runs on your infrastructure.
- •
Data and processing-heavy platforms, where cost scales with volume regardless of seat count.
- •
Installed or on-premise products that call vendor-hosted AI services, which quietly turn a fixed license into a variable cost.
It matters much less for desktop, on-premise or air-gapped software where the customer's own hardware does the work. For those vendors, usage-based options are about capturing value and giving customers flexibility, not about recovering costs.
Usage pricing can still help margins indirectly. Some customers won't sign a large fixed fee up front, and letting them start small and grow turns a lost deal into a land-and-expand account whose margins improve as it scales.
If usage doesn't map to value in your product, you can protect margin in other ways: fair-use limits inside existing plans, a priced AI add-on, or tiering that moves heavy workloads into a higher package. Credit-based pricing lets you put a price on the expensive capability without repricing the whole product.
The decision tree
Oliver Wyman's October 2026 analysis asks whether usage pricing is feasible for a product and whether it would beat the economics the vendor has today. Its survey of more than 150 North American software buyers found close to two-thirds view usage pricing favorably, while roughly nine in ten worry about unpredictable costs. We've turned those tests into a sequence. Work through the questions in order, stop at the first "no," and take that branch.
Step 1: Can you name what's driving the change?
If the only honest answer is "investors" or "everyone else is doing it," talk to customers before you touch price.
Step 2: Does a usage metric track the value customers receive?
If not, keep subscription or seat pricing and repackage. Protect margin with fair-use limits or a priced add-on for cost-heavy features.
Step 3: Do your cost drivers move with that metric?
If not, split the offer. Keep the core on subscription and put the expensive capability on credits or a metered add-on.
Step 4: Can customers forecast their bill and keep it in check?
If not, use a hybrid pricing model with a subscription floor. The eight pricing shapes customers will buy show how caps, included usage and adaptive flat rates give buyers a number they can plan around.
Step 5: Can you meter usage, enforce limits and change prices without an engineering release?
If not, fix the infrastructure before you change the price.
Step 6: If every answer is yes
Test usage-based pricing with a cohort. Start with new customers or an opt-in plan rather than a forced migration.
Most established B2B vendors stop at step 4. Poyar's report found 76% of AI-native companies already use some form of variable or hybrid pricing, against 49% of SaaS companies. Gartner, as reported by CIO.com, predicts at least 40% of enterprise SaaS spending will shift toward usage-, agent- or outcome-based pricing by 2030. If you want a head-to-head on predictability and expansion, read usage-based pricing vs subscription.
Your customers aren't asking. Should you worry?
Plenty of B2B vendors look at their renewal data, see no pressure, and decide they can wait. Sometimes they're right. Oliver Wyman found few enterprise buyers expect seat counts to fall faster over the next three years than over the last three, and it called the current moment a window to act rather than a reprieve.
Customers rarely ask for a pricing model nobody has offered them, but they do switch when a competitor offers one. AI-native challengers now enter established categories charging per resolved ticket or completed task. A buyer comparing one of them with a per-seat incumbent is weighing a fixed cost against a cost that only arrives with results.
You'll see this first in lost deals and in procurement asking for "something more flexible," long before it turns up in churn.
So if you're moving slowly, ask how long it would take you to respond if a competitor repriced this quarter. If the honest answer is a multi-quarter engineering project, that delay is your exposure.
How fast you can respond comes down to where your pricing lives. AI features ship monthly, inference costs shift between models, and competitors repackage between your renewal cycles, so pricing and packaging can't wait for an annual review anymore. The vendors who learn fastest run small tests all the time, and they can only do that when plans, limits and entitlements live outside the code. A monetization layer that sits apart from the product and the engineering roadmap makes a pricing change a routine task. We cover how that works in packaging and pricing as configuration, not code.
The packaging questions that matter more than the model
The usage-versus-seats debate gets the attention, but the choices below decide whether your new pricing makes money, and you'll need data for each one.
Functionality you don't control today
Many established products have capabilities that were never gated. They shipped to everyone, and nobody tracks who uses them. You can't price, bundle or limit a capability until there's an entitlement around it, so start by listing the features that run with no entitlement check and decide which of them carry real value or delivery cost.
New functionality: charge for it or give it away?
Not every new feature should have a price. Some exist to drive adoption of the core product, defend renewals or answer a competitor. Others carry delivery cost or create clear new value, and they deserve a price, an add-on or a usage allowance. Ask whether the feature changes what a customer would pay or whether they stay. Charge for the first kind, and consider including the second kind in your existing plans.
Old cohorts versus new cohorts
New pricing for new customers is the lowest-risk test you can run. Existing customers can stay on their current terms, move by choice, or move at renewal. If you offer the new plan alongside existing subscriptions instead of forcing a switch, you also get a comparison group. Running two cohorts on different terms only works if your systems can hold both without manual workarounds.
Renewal timing
Renewal is the natural moment to change terms. Map who renews in which quarter and sequence your changes so you learn from the early cohorts before the large accounts come up.
Forced upgrades versus optional ones
Forced migrations are quick to administer, and they're where pricing changes go wrong in public. Optional migrations take longer but show you who moves and why. A middle path makes the new plan the default at renewal and lets customers who ask keep their current terms for a set period.
Measuring uptake per package and per customer
Every one of these decisions depends on knowing how customers use what they bought: which features in each package get used, which accounts are near their limits, and which never adopted the capability you bundled in. Without that, packaging changes are guesses. Revenue Intelligence shows uptake and engagement by package and by account, so each decision rests on what the last test showed.
The operational work nobody budgets for
Pricing strategy fails most often in operations. These are the questions that break SaaS and licensing teams:
- •
How do you add functionality to an existing plan for every customer on it, automatically, without touching each account?
- •
How do you move customers to a new plan, at the same price or a different one, and keep their custom terms intact?
- •
How do you track who had which entitlements, when, and why, across years of changes?
- •
How do you handle custom entitlements for strategic accounts without creating a one-off product per customer?
- •
Where does the record of entitlements live?
Many companies answer that last one with their CRM. Salesforce knows what was sold. It doesn't know what the customer is allowed to run, what they've consumed, or what the product should enforce right now. That's the job of an entitlement database, and it has to stay in sync with CRM, billing and ERP.
Where that system doesn't exist, the symptoms are familiar. Pricing is fixed in the code. Workflows break between teams because CRM, billing and the product don't talk to each other. Finance reconciles usage and invoices by hand, and errors slip through every month. Usually nobody budgeted for a proper back end, because licensing was treated as a one-time build instead of a revenue system.
Fixing this is less glamorous than choosing a pricing model, and it's worth more. A single monetization infrastructure that holds entitlements, enforces them in the product and feeds usage to billing removes most of the manual reconciliation and turns every change above into a configuration task.
A practical path for legacy B2B vendors
If you have an installed base, perpetual licenses or long-standing subscriptions, the order you do things in matters more than how fast you do them.
Separate pricing from the product. Move entitlements and plan logic out of the codebase and into a system your product and revenue teams can change. Licensing & Entitlements covers installed, on-premise and offline software, and the Monetization Engine covers hosted SaaS and AI services.
Meter before you charge. Track usage for several months with no price attached. You'll learn which metric tracks value and how much customers vary. The Zenmeter documentation covers metering, rate cards and usage caps.
Pilot with a defined cohort. Use new customers, one region or an opt-in plan. Keep a subscription floor and add included usage.
Build guardrails before launch. Give customers caps, alerts, committed drawdowns and real-time visibility into their usage. The usage-based billing FAQ covers the questions buyers raise most.
Align compensation. Pay reps on new logos and consumption as well as bookings. Reps paid only on bookings will oversell the new model or avoid it.
Review monthly. Use uptake data to adjust allowances, tiers and add-ons in small steps.
Preserve how you sell software today. Unlock how you'll monetize it tomorrow.
Homegrown and legacy licensing systems were built around perpetual licenses, seats and fixed entitlements, so every move toward usage-based models turns into an engineering project.
The Nalpeiron Growth Platform replaces that infrastructure with one platform for licensing, entitlement enforcement, granular usage metering and monetization. Your developers and product managers keep supporting existing customers and licensing models while they introduce subscriptions, consumption-based pricing and hybrid offerings, without rebuilding the licensing layer each time. Nalpeiron controls rights, limits and usage, then passes rated usage to the billing system or ERP you already run.
Modernize your licensing once, protect your existing business, and gain the freedom to evolve pricing and packaging as the market moves.
Frequently Asked Questions
Should a legacy B2B software company move to usage-based pricing?
Only if usage tracks the value customers receive, delivery cost rises with it, and buyers can budget for the bill. Most established vendors meet the first condition but not the third, which points to adding usage on top of a subscription.
Is usage-based pricing a fix for AI margin compression?
Not automatically. Margin compression usually means your price and your cost to serve now follow different metrics. Usage pricing helps only if usage also maps to value. If it doesn't, fair-use limits, priced AI add-ons or credits for cost-heavy features protect margin without a full model change.
Who is most affected by AI margin compression?
Vendors that pay for the compute. That includes hosted SaaS and AI products, data-heavy platforms, and installed software that depends on hosted AI. Software running on customer hardware is far less exposed.
What if customers aren't asking for usage-based pricing?
Buyers seldom request an option they've never seen, so the pressure usually arrives through a competitor that reprices the category. Measure how quickly you could respond, which depends on whether pricing can change without an engineering release.
Should new features be charged for or included in existing plans?
Charge for a feature when it raises willingness to pay or adds meaningful delivery cost. Include it when its main job is adoption, retention or competitive defense. Usage data on comparable features is the best guide.
Why can't the CRM be the system of record for entitlements?
A CRM records what was sold. An entitlement system holds each account's rights, consumption and live enforcement rules. The two need to stay in sync, but they do different jobs.
Nalpeiron: A Long-Term Partner for the AI Era
At Nalpeiron, we go beyond technology — we act as a strategic partner in licensing, monetization, and growth. For over twenty years, enterprise and IoT companies have trusted us to guide and evolve their business models.
As AI shifts software from seats to usage, outcomes, and agent-driven activity, legacy approaches fall short. Nalpeiron enables this transition through entitlements as the control plane — a centralized system of record across SaaS, on-prem, IoT, and offline environments.
From strategy to execution, we help companies adapt faster, launch new models, and stay in control — making Nalpeiron a partner for the AI-driven future of software monetization.
Ready to Optimize Your Strategy?
See how Nalpeiron helps companies implement flexible monetization strategies that support both product-led and sales-led growth motions.
Book a Demo