## TL;DR

The OCR picked up the wrong number (subtotal, tax, or a nearby figure) as the total. Overwrite the total with the correct figure from the receipt, save, and verify the correction persists; if the tool re-scans and overwrites your fix, disable auto re-scan for that expense.

## Error

```text
receipt ocr misread the total: correction failed
```

## Steps

1. Open the receipt image and read the actual total yourself, checking whether the OCR grabbed the subtotal, the tax line, or a loyalty-points number. Expected: you know the right figure and what confused the OCR.
2. Edit the expense total to the correct figure and save. Expected: the expense shows the corrected total.
3. Reopen the expense to confirm the correction stuck. Expected: the total is still your corrected figure.
4. If the tool reverted your fix, look for a re-scan or refresh option on the expense and avoid triggering it; enter any remaining fields manually. Expected: your correction survives.
5. For vendors whose receipts routinely confuse the OCR, note the pattern for your team. Expected: reviewers know to double-check totals from that vendor.

## When to use

- The OCR total is wrong (subtotal, tax, or stray number captured).
- Your manual correction gets overwritten or does not save.

## When not to use

- The merchant name is wrong but the total is right (see the merchant skill).
- The whole scan failed (see the unable-to-read-receipt skill).

## Tool compatibility

- Expensify SmartScan, Concur, and other OCR-based receipt capture.

## Variant phrasings

### Incorrect total extracted from receipt

Same misread, alternate wording.

### OCR picked up the wrong amount

The plain-language form of the same problem.

## Why it happens

Receipt layouts vary: the total is not always the largest or last number. OCR picks the most total-like figure, and similar numbers (subtotal, balance due, points) fool it.

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst__B30d1WVIXcZcWkFDfpKcA