What is usage-based pricing — and when should you switch from per-seat?

Direct Answer

Usage-based pricing (UBP) means the customer pays for consumption — API calls, GB stored, events, transactions, host-hours — rather than per-seat licenses. Snowflake bills credits, AWS bills every dimension, Twilio bills messages, Datadog bills host-hours plus events, Stripe takes a percentage.

Switch from per-seat to UBP (usually hybrid, not pure usage) when buyer value scales with consumption rather than headcount, when seat counts are a poor proxy for what your product does, or when seat-license friction throttles land-and-expand inside accounts.

TL;DR

• UBP charges for consumption, not headcount; ~46% of cloud-native SaaS now has some usage component in 2027, up from 26% in 2020 (OpenView 2024 PLG Index).
• Pure-usage suits ~15% of B2B SaaS (infra, API, data); pure-seat suits ~35% (workflow); the other 50% should run hybrid — seat base plus usage on one metered dimension.
• Switch when value scales with consumption, when seat-count is a proxy mismatch, or when license friction is blocking champion-led expansion.
• Do not switch for workflow products like Slack or Salesforce CRM; do not switch without spend caps and forecast tools — surprise bills churn customers.
• Three killers of UBP rollouts: no spend caps, sales comp not rewired, and broken metering you cannot reconcile against invoices.

The 3 Pricing Models and When Each Wins

The 2027 landscape has settled into three archetypes. The mistake most RevOps teams make is forcing one model onto a product that wants another.

The honest read: pure usage is right for ~15% of B2B SaaS (infrastructure, API, data). Pure seat is right for ~35% (workflow tools where a person does the work). The remaining 50% should run hybrid — the Datadog template: a per-host base finance can forecast, plus a per-event meter that captures upside when telemetry volume explodes.

What Changes When You Switch

Switching pricing is an operating-model project, not a billing project. Four functions move in lockstep.

Sales comp gets rewired first. If AEs sell a pre-paid commit but get paid only on first-year ACV, they sandbag commits and the flywheel never spins. 2027 best practice: pay AEs on the commit at booking, with a true-up accelerator on overage realized in the same fiscal year.

Finance forecasting changes shape. ARR becomes "committed ARR plus run-rate consumption," and guidance has to acknowledge consumption can swing 15-25% quarter to quarter on macro alone. Snowflake guides product revenue, not ARR, because consumption-led businesses cannot pretend monthly run-rate is locked.

Customer success flips from renewal-led to consumption-led. CSMs own the burn-down curve: if a customer is 60% through the quarter and only 30% through their commit, that is a save motion. The playbook becomes weekly consumption reviews and expansion triggered at 70-80% burn — not annual QBRs.

Billing infrastructure is the unglamorous foundation: reliable event ingestion, idempotent metering, and a usage ledger finance can reconcile against invoice lines. This is why Metronome, Lago, Stripe Billing with Usage Records, Maxio (Chargify+SaaSOptics merger), and Subskribe's native usage CPQ exist — building it yourself is a multi-engineer-year distraction.

The 3 Implementation Failures

The first killer is launching usage without spend caps and alerts. A customer signs a pre-paid commit, a misconfigured pipeline emits ten times the events, and four weeks later they get a surprise $400K overage invoice. They do not pay, they churn, and NRR takes a hit.

Every UBP rollout needs hard caps, soft caps with alerts at 50/80/100% of commit, and a self-serve dashboard.

The second killer is sales comp that has not caught up. AEs who sold seats for a decade will keep selling seats — quoting "estimated usage" as a seat count, the customer under-commits, expansion never materializes. Fix this before launch, not after the first miss.

The third killer is metering you cannot trust. If engineering cannot reconcile billed events against emitted events, you cannot defend an invoice when a customer disputes it. Treat metering as a tier-zero system: SLAs, dual-write reconciliation, quarterly audit against product telemetry.

Frequently Asked Questions

Metronome vs Lago in 2027 — which should we pick? Metronome ($30-150K/yr) is the enterprise pick: battle-tested at OpenAI and Confluent scale, with mature commit-and-true-up logic. Lago is the open-source challenger — right when you have engineering capacity, need self-host for data-residency, or the $30K floor matters.

For most Series B+ companies, Metronome is the faster path to a defensible invoice.

Should AEs sell pre-paid commits or month-to-month usage? Pre-paid commits, almost always. Commits give finance the bookings number, give the customer a discount for forecast certainty, give the AE a clean comp event. Month-to-month works for self-serve PLG, but enterprise needs a commit anchor — Snowflake, Datadog, Twilio all run this way.

How do customers actually budget on UBP? Pre-paid annual commit, a forecast view inside the product, procurement-approved soft caps that trigger internal review before overage. UBP winners make budgeting easier than per-seat.

Sources

• OpenView Partners, 2024 PLG Index — UBP adoption rose from 26% (2020) to ~46% (2024) of cloud-native SaaS.
• Bessemer Venture Partners, State of the Cloud 2024 — hybrid dominant; pure-usage ~12%, pure-seat ~35%.
• Tomasz Tunguz, "The Hybrid Pricing Era" (2024) — seat+usage convergence across public SaaS.
• Metronome, 2024 UBP Benchmark Report — commit-and-true-up patterns, AE comp structures.
• Snowflake FY2024 Q4 earnings call — consumption guidance methodology; product revenue vs ARR.
• Datadog 2024 Investor Day — host-plus-event hybrid model and net retention dynamics.
• Pavilion 2024 Pricing and Packaging Survey — 800+ RevOps leaders on post-UBP comp rewiring.
• A16z, "The New Business of Consumption" (2024) — finance/procurement implications.