Metered billing

Metered billing bills a customer for the units a meter recorded in the period — API requests, tokens, gigabytes, minutes, messages — at an agreed rate. “Metered” names the mechanism; usage-based billing names the pricing model built on it. In practice the terms are used interchangeably.

How a meter works

• A metric defines what is counted and how: sum of quantities, count of events, count unique of a key (active users), max, min or last value in the period.
• Events arrive from the product with a quantity, a timestamp and an idempotency key, so a retried event never counts twice.
• Late events — ones that arrive after their period — must land in the period they belong to, or be corrected on the next invoice.
• Refused events (unknown metric, unknown customer) need somewhere to go other than silently disappearing.

Why it matters

The meter is the revenue record. A meter that drops 1% of events underbills by 1%, invisibly, every month. A meter that double-counts produces the overage dispute. Accuracy — deduplication, late-arrival handling and a queue for refused events — matters more than pricing flexibility.

How Fynex does it

Fynex’s metering accepts events one at a time, in batches of up to 500, by CSV or through signed inbound webhooks, deduplicates on the idempotency key, places late events in their period, and holds refused events in a dead-letter queue for redrive. See usage-based billing.