The Analyst's Path

Usage-based pricing model

With metering, the customer pays in proportion to consumption: per GB, per kWh traded, per API call, per ticket booked, per compute-second. No other structure tracks delivered value as closely, and that cuts both ways. Adoption friction is low, because there's no big commitment. Expansion is automatic too. Usage grows, so revenue grows, so net revenue retention runs above 100% with zero selling effort. But revenue is volatile by design. In a downturn customers can optimise, throttle, and cut consumption instantly, and no contract stops them.

2 min readPart of Business Model AnatomyOpen in course mode

Advertisement

With metering, the customer pays in proportion to consumption: per GB, per kWh traded, per API call, per ticket booked, per compute-second. No other structure tracks delivered value as closely, and that cuts both ways. Adoption friction is low, because there's no big commitment. Expansion is automatic too. Usage grows, so revenue grows, so net revenue retention runs above 100% with zero selling effort. But revenue is volatile by design. In a downturn customers can optimise, throttle, and cut consumption instantly, and no contract stops them. Cloud's 2022 to 23 "optimisation" quarters are the canonical demonstration, with AWS growth halving in a year and Snowflake's consumption dipping. Subscription revenue would have held. Metered revenue repriced to reality in real time.

Usage-based pricing model examples

  • Snowflake (US, illustrative ~FY25): product revenue ~$3.5bn (+~30%), NRR ~126% and gross margin ~75%, yet it reports a GAAP loss, because stock-based compensation runs ~40% of revenue. It's your cleanest live specimen of the SBC leak in the waterfall. Claims of an "FCF margin ~26%" that add back SBC are counting dilution as profit.
  • AWS (within Amazon): ~$108bn revenue and ~$40bn operating income, from usage pricing on top of staggering fixed capex. The meter is high-margin only because the underlying data centres are highly utilised. That's operating leverage again.
  • India: IEX collects ~₹0.02/kWh from each side of every electricity trade. That's ~₹550 to 600 cr of revenue at ~60%+ PAT margins, near-perfect toll economics with one master risk: regulation of the toll itself, as in the market-coupling proposals. IRCTC's convenience fee of ₹15 to 30 per online ticket has the same anatomy and the same risk. Worked Example 4 covers the day the government touched the meter. Telecom prepaid, such as an Airtel recharge, is metered access sold in sachets, India's oldest usage model.

Advertisement

What should you check in the filings?

What to check in filings: consumption against commitments. Cloud contracts often mix committed minimums + overage, and the RPO tells you the floor. NRR and its decomposition. Unit pricing trends matter because per-GB and per-call prices always deflate, so volume has to outrun them. Utilisation of the fixed asset behind the meter. And where the meter is an essential facility, the regulator's hand.

Advertisement

Part of the course

This page is part of Business Model Anatomy (M4.01), in Phase 4: Business models, competitive strategy and moats. The module's flashcards and its quiz are in the app.