Start with the right question
A scanned document can look correct while its hidden text contains mistakes. Those mistakes can flow into search, summaries, and calculations. Review the fields that matter to your task and keep a path back to the visible original page.
Work through the checks
Identify the source quality
Record missing pages, skewed scans, faint print, handwriting, and cropped edges. A clean extraction cannot compensate for material absent from the image. Request a better copy where important text is unreadable.
Check numbers and signs
Review decimal points, minus signs, dates, quantities, and currency symbols. Similar-looking characters can change an amount materially. Recalculate totals from verified values rather than trusting an extracted table.
Inspect table relationships
Confirm that values remain attached to the correct row and heading. Multi-column layouts and page breaks can move amounts beside the wrong description even when each number was recognized correctly.
Keep corrections traceable
Record the extracted value, verified value, and page reference. Do not overwrite the original scan. Mark uncertain characters instead of choosing the value that produces the preferred result.
A record you can use
- Source page
- Extracted field
- Visible original value
- Correction or uncertainty
- Reviewer
Keep this record locally. Use document references rather than publishing account numbers or other people's personal details. An empty field means the information is still missing; it should not be filled with an AI guess.
Worked example
A fictional scan shows $18.50, but extracted text reads $1850. The reviewer checks the image and corrects the ledger before a total is calculated. A second row remains unreadable and is flagged for a better copy rather than guessed from the invoice subtotal.
This example is fictional and illustrates the method, not a real case result.
Limits and next steps
OCR verification does not establish the truth of the underlying document or whether a cost is recoverable. Important quotations and formal submissions may require additional review beyond a text extraction check.
Use verified fields in summaries and calculations and retain the correction table. If an AI tool reads the document, make sure it receives the corrected context without altering the original evidence.