Migrating Customers From Seat-Based to Credit-Based Pricing

Migrating customers from seat-based to credit-based pricing is a trust problem as much as a billing change. The transition succeeds when customers can understand what credits buy, predict their spending, and move without a surprise invoice. That requires a structured customer journey, built around clear value mappings, grandfathered transitions, and real-time consumption visibility.

Why seat-based pricing breaks down when AI workloads enter the picture

A seat charges for access by a named human. An AI agent consumes compute cycles, tokens, and API calls that bear no consistent relationship to headcount. As agents take on more work, a customer can reduce its seats while increasing its software consumption. Revenue then moves against the product’s growing usefulness. The mismatch compounds as more activity shifts from people to agents.

Customers pay for seats the agents do not use and for the compute they consume. The financial reason to act is the net revenue retention gap, as usage-priced companies retain a materially higher share of revenue year over year than seat-priced peers.

Why credits became the dominant transition mechanism rather than pure consumption pricing

Credits give consumption a budget customers can approve. A vendor sells a balance in advance, and product actions draw it down at defined rates. The buyer commits to capacity while retaining flexibility over its use. Customers can therefore treat credits as budget units rather than separate cash outflows at every action, while vendors retain committed revenue and expansion potential.

Pure pay-as-you-go leaves buyers anxious about an open-ended invoice, while outcome-based pricing remains difficult to define contractually. Bundled allowances such as Atlassian’s Rovo credits illustrate this middle ground without requiring an immediate move to unrestricted consumption billing.

The commitment itself is not the central design risk. Buyers gain a predictable budget only if rollover and overage policies preserve that predictability. The exchange rate must also be intelligible, ideally linking a credit to a recognizable enrichment, conversation, or generation. Otherwise, the private currency merely conceals the uncertainty it was meant to resolve.

What the actual transition looks like — five vendors that rebuilt billing in 2025–2026

monday.com chose a seat-plus-credits model on May 6, 2026, with a matching monthly discount offsetting each bundle. No customer’s bill increased on transition day. That design separated the introduction of credits from an immediate price increase, leaving expansion revenue to consumption beyond the bundle.

GitHub Copilot replaced premium request units with AI Credits on June 1, 2026. Those credits are used up as tokens are consumed, at published API rates. Prices stayed the same, but what they bought changed. Access became bounded and variable rather than unlimited and flat, connecting the allowance to the resources consumed.

Docusign took a less common path when it removed seats from Enterprise IAM in June 2026. Any number of users could draw from a credit bucket shared across departments. The action replaced the user as the billing unit. For a cross-department workflow product, a signature or workflow execution provides a more direct basis for charging than the number of people with access.

HubSpot layered Breeze AI credits, priced at $9 per 1,000, onto its existing seat structure. In April 2026, it moved Breeze Customer Agent from $1.00 per conversation to $0.50 per resolved conversation. Payment shifted toward outcomes the AI completed. The change also illustrates how revising a metric mid-product cycle creates guidance uncertainty, even when the new unit better reflects value.

Atlassian first sold Rovo as a monthly AI add-on, then bundled credit allotments into Jira and Confluence tiers at no additional cost. Adoption preceded enforcement of paid overages. Overage billing takes effect on December 3, 2026, with extra usage enabled by default and no billing for credits above the allowance before that date. Its 90-day notice policy gives customers time to adjust before enforcement, making advance warning part of the migration design.

Segmenting the customer base before a single migration message goes out

Segment customers by the financial effect of migration. Winners pay less under credits, neutral accounts pay roughly the same, and high-impact accounts pay materially more. Each group needs a different conversation. In particular, heavy users with relatively few seats need early personal engagement.

Shadow billing supplies the evidence for those conversations. Run metering, aggregation, and invoice generation in parallel with live seat billing before sending a new invoice. Compare those shadow invoices line by line with what customers pay now. This is not merely an engineering rehearsal. It establishes what each customer would actually pay and prevents the segmentation from resting on assumptions that later become surprise bills.

New Relic’s move from host-based subscriptions to pure consumption billing illustrates why cost alignment with value is more important than migration speed. The lesson for segmentation is to make that alignment demonstrable account by account. A projected bill must explain its own logic. If it cannot show in plain language what credits buy for each use case, the credit-to-value mapping is not ready to publish.

Defining the credit-to-value mapping customers will actually trust

Start with an intuition test before finalizing the exchange rate. A customer should be able to determine an action’s credit cost before executing it. Publish an explicit exchange table linking product actions to credits, with worked examples for the two or three most common workflows. Customer success teams need that same document when they talk to customers about migration. Abstract units without recognizable outcomes leave buyers unable to evaluate the commitment.

Next, define how the balance behaves across the account. Pooling should make unused credits movable between users or departments, removing the allocation friction of seats. Rollover should let customers use their unused credits in the next period. Without it, lower consumption can feel like a penalty.

Cost controls must be equally explicit before launch. Show the credit cost before a customer executes an action, especially for expensive workflows. Set a hard overage cap so that spending above it needs explicit approval. That removes open-ended liability from the commitment. Together, the preview and cap make the published exchange table usable at the point of decision.

Salesforce’s Agentforce trajectory provides a caution about unsettled mappings. It launched with a flat per-conversation rate. After enterprise pushback, it expanded to options including per-action flex credits and per-user licensing.

How to structure the customer communication timeline — and what to say to each segment

Once segmentation and the exchange table are ready, launch the new architecture with new customers first. Existing customers should stay on seat billing under a grandfather arrangement while the model proves itself. This separates product risk from migration risk. Enterprise accounts also need notice well ahead of a change so that budget cycles do not become forced escalations. Atlassian’s 90-day quota-enforcement notice provides a minimum runway for accounts accustomed to an unenforced allowance.

Use the shadow invoices to decide which existing customers to talk to first. Winners should hear first in a conversation showing their actual usage and what they would have paid. Neutral accounts need a written explanation, the exchange table, a migration date, and a named customer success contact. High-impact accounts need a meeting of their own before they get written notice. The purpose is to co-design their migration path.

Choose the mid-cycle treatment before communicating a date. One option is to credit the unused portion of the seat-based cycle and begin credit billing on transition day. The alternative is to keep seat billing through the current cycle and switch at renewal. The former is cleaner for accounting but needs explanation. The latter is easier to communicate but delays migration. monday.com’s bill-neutral launch shows why the financial effect on transition day deserves an explicit commitment.

Lead the message with evidence showing what the customer used, what that usage would cost, and how the transition protects the budget. Protection needs a mechanism. After launch, the product must make those same facts continuously visible.

Real-time spend visibility as the infrastructure that keeps the new model trusted after launch

A credit dashboard must show how much customers have used while they still have time to act. Its core elements are a current balance, previews before high-cost actions, threshold alerts before exhaustion, and line items expressed as product actions. Raw event counts do not explain a bill. The dashboard turns the exchange table into something customers can observe, rather than a document they must trust without evidence.

That experience depends on durable event ingestion and idempotency. A lost event removes revenue and part of the invoice’s audit trail, while a duplicated retry must not deduct credits twice. Batch aggregation adds risk when balances take hours to reflect what customers have used. Alerts then arrive too late to protect the budget.

A separate metering tool feeding a separate billing system creates a seam where lag can accumulate and audit trails can fracture. A unified engine for usage ingestion and invoice generation removes that particular handoff. It does not replace sound event design. Purpose-built billing infrastructure is useful because it links downstream billing to the record of what customers used, which they need to understand their charges.

Own the event contract and billable-unit definitions because the value metric is closely tied to the product. Invoice generation, dunning, tax, and payment methods do not require the same reinvention. Use infrastructure built for those downstream tasks.

Protecting revenue during the transition — grandfathering, pricing floors, and the NRR argument for moving

For finance leaders, migration creates a bounded short-term exposure. Churn risk is highest during the transition, when customers must reassess both their spending and their trust in the vendor. Segmentation and early engagement address that exposure directly. Grandfathering buys time, but it needs a defined end. Tie expiration to renewal rather than an arbitrary calendar date so procurement can evaluate the change at a natural decision point.

Credit rates also need a delivery-cost floor. AI inference costs vary with usage, so expansion is not automatically profitable. ICONIQ Capital’s 2026 State of AI report places AI-feature gross margins materially below traditional SaaS margins while emphasizing rapid improvement. A model that aligns charges with consumption must still cover the cost of serving it.

Those transition costs come with a chance to keep more revenue over time. Usage-based pricing allows revenue to expand with consumption independently of growth in the number of human users. It can also reduce the resentment of low-utilization customers paying for unused capacity. That is the operating basis for the directional NRR advantage. The board’s question is whether the new model converts better alignment into durable revenue without sacrificing delivery economics.

The sequence should make that case with evidence. Prove the model on new customers, then grandfather the existing base with a defined expiration. Let the new-customer cohort’s NRR trend support any acceleration of the existing-customer migration. The first step is therefore not a public announcement, but the analysis that makes a defensible customer offer possible.