Pay-per-use charges for actual consumption: per kilowatt hour, per API call, per gigabyte, per hour of rental, per printed page. The customer pays nothing when they do not use the service. Your revenue moves with their activity, and your costs usually move with it too, which keeps margins stable at any volume.
Cloud computing and car sharing are the familiar examples. You pay for the minutes or the gigabytes you actually used.
The attraction is margin stability. Suppose you charge 0.10 euros per 1,000 API calls and your infrastructure costs 0.04 euros per 1,000 calls. A customer with 2 million calls a month pays 200 euros and costs you 80 euros, leaving 120 euros. A customer with 200,000 calls pays 20 euros and costs 8 euros, leaving 12 euros. In both cases you keep 60 percent. Under a flat rate of 150 euros a month, the first customer would produce a loss and the second a large profit.
The weakness is predictability. You have no revenue floor. If your customers have a quiet quarter, you have a quiet quarter, and the rent stays the same. Many providers therefore combine models: a small base fee plus usage, or a minimum monthly commitment with usage counted against it.
Customers have the opposite fear, an invoice they cannot forecast. Reduce it with a spending cap, an alert at 80 percent of an agreed limit, and a live usage display. One unexpected four-figure bill ends a relationship faster than a price increase.
Metering must be exact and visible. If a customer cannot reproduce how the invoice was calculated, they will dispute it. Log every unit, show the log, and define in the contract what counts as one unit: a failed API call, a cancelled booking, a partial hour.
Pay-per-use appears in generated business plans as one of the evaluated revenue patterns, usually opposite the flat rate. The choice depends on whether your costs rise with usage.
