Why copying a PDF table often fails
A PDF is designed to preserve page appearance, not spreadsheet structure. Even when text is selectable, copied values may arrive in one column, lose row boundaries, or mix headers with data. Scanned PDFs have an additional problem: there is no text layer to copy at all.
Table OCR looks at the visual layout and attempts to rebuild cells, rows, and columns. The result still deserves review, but it gives you a structured starting point.
Step-by-step workflow
- Choose only the useful pages. Smaller, focused files are easier to review. Remove cover pages and unrelated appendices when possible.
- Check table quality. Straight pages, clear column spacing, and legible numbers matter more than decorative borders.
- Select Sheets output. Choose spreadsheet or table mode in the OCRToDocs dashboard or launch the add-on from Google Sheets.
- Run OCR and inspect the preview. Confirm that headers, quantities, currency values, and totals occupy the expected columns.
- Clean number formats. Convert imported currency, date, and percentage text into Google Sheets number formats before calculating totals.
What to review after extraction
| Check | Why it matters |
|---|---|
| Header row | A shifted header can make every following column ambiguous. |
| Decimal separators | Periods and commas vary by locale and can change numeric values. |
| Negative values | Parentheses, minus signs, and credits may be missed in faint scans. |
| Merged cells | Multi-line labels can create extra columns or rows. |
| Grand totals | Compare extracted totals with the source before relying on formulas. |
When to use document output instead
If the PDF is mostly paragraphs, contracts, or reports with only one small table, document OCR may preserve the reading flow better. Use Google Docs OCR for narrative text and Google Sheets OCR when rows and columns are the main value.