## TL;DR

The upcoming-invoice endpoint previews what a metered-billing customer owes right now, but it is an estimate: usage keeps accruing until the cycle closes, and the final invoice can differ. Use it for in-app 'current usage' displays and for support answering 'what will I owe', with a clear 'as of now' label. Never present it as the final amount or use it for accounting. Cache it briefly rather than calling per page view; usage previews do not need real-time precision.

## The query

```text
Stripe upcoming invoice for metered billing: estimates vs final amounts
```

## Use this when

- Showing customers their current metered usage cost
- Support answering what will I owe questions
- Previewing metered invoices before cycle close

## Not for

- Finalized invoice amounts (use the invoice object)
- Revenue recognition (never on estimates)

## Steps

1. Call the upcoming-invoice endpoint for the metered customer.
   Expected output: You get the current estimate with usage lines.
2. Label it clearly as an estimate as of the current time.
   Expected output: Nobody mistakes it for the final bill.
3. Display it in-app or hand it to support with the as-of timestamp.
   Expected output: Customers see a useful preview.
4. At cycle close, compare the estimate logic against the finalized invoice.
   Expected output: Your preview math stays calibrated.
5. Cache the result for minutes, not seconds, to avoid API spam.
   Expected output: The preview stays cheap to serve.

## Variant phrasings

### Stripe upcoming invoice metered billing estimate

### preview usage based invoice Stripe

### upcoming invoice vs finalized invoice Stripe

## Root cause

The upcoming invoice is computed from usage reported so far, and usage is a stream that does not stop until the period closes; the estimate is therefore a snapshot, not a promise. Treating it as final is how you get 'but you told me X' disputes.

## Edge cases

- Discounts and coupons preview too, but their final application follows the same order rules
- Very high usage volumes can make the preview slow; cache aggressively

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_uHDieFbVM2WoL3lZR80_mA