How does column detection reconstruct tables from PDFs?
The converter groups text runs onto shared horizontal baselines and splits columns wherever gaps between adjacent words exceed 1.2 times the font size. Under the Library of Congress PDF 1.7 (ISO 32000-1) Format Description, PDFs store positioned glyphs in 1/72 inch coordinates rather than native spreadsheet rows and columns.
Grouping baselines and horizontal column gaps
Even in untagged PDF documents that contain no structural table tags, table columns are detected directly from visual coordinates. The parser sorts every visible text fragment vertically and clusters items sharing a baseline within 2 pt or 45 percent of the font height, then starts a new cell whenever horizontal whitespace exceeds 1.2 times the character size.
Column starting positions across the entire page are then clustered within a 12 pt window (using a 6 pt left-edge tolerance). Each cell snaps into the closest aligned column, turning raw PDF coordinates into clean rows without requiring manual gridlines.
Handling scanned statements with automatic local OCR
When a bank statement or receipt arrives as a flat image scan with no selectable text layer, the tool triggers local OCR automatically page by page. If you want a searchable text-layered PDF copy alongside your spreadsheet, run the scan through a browser-based OCR scanner as well.
Turning formatted amounts into real spreadsheet numbers
Leaving Turn amounts into numbers checked converts formatted currency and percentage strings into genuine numeric values instead of plain text labels. Many online converters leave values like $1,250.00 or (45.50) as text strings, causing Excel SUM formulas to return zero until you strip symbols by hand.
Parsing US, European, currency, and accounting negatives
| Input PDF cell text | Parsed Excel value | Numeric rule applied |
|---|---|---|
1,250.00 |
1250 |
Strips US thousands commas |
1.234,56 |
1234.56 |
Converts European decimal comma |
(45.50) |
-45.5 |
Turns accounting parentheses negative |
€12 or $99.50 |
12 or 99.5 |
Strips currency symbol prefix |
12% |
0.12 |
Divides percentage by 100 |
Currency prefixes ($, €, £, ¥, ₹), minus signs, and accounting parentheses all normalize cleanly into IEEE numbers. During extraction, European decimal commas (such as 1.234,56 or 45,50) are detected when a comma precedes one or two trailing digits, stripping thousands dots and converting the comma into a standard decimal point.
Uncheck Turn amounts into numbers only if you need to preserve leading zeros in ZIP codes or reference IDs that look like pure numbers.
Building standard ECMA-376 SpreadsheetML .xlsx files
Following the Ecma International ECMA-376 Office Open XML Standard, the writer emits numeric <v> elements without string-type flags for parsed amounts while escaping XML 1.0 control characters. Once your figures are edited, you can export spreadsheets back into PDF tables for clean sharing.
Client-side PDF extraction avoids the file size queues and retention risks of cloud converters, letting you convert multi-page quarterly statements instantly in browser memory.
When should you use one sheet per page versus a single sheet?
Choose One sheet per page when each page of a report has different column widths, and pick All pages on one sheet when extracting a continuous multi-page ledger. Under the Library of Congress XLSX (OOXML SpreadsheetML) Format Description, a single .xlsx ZIP archive can hold multiple named worksheets (Page 1, Page 2, and so on).
| Sheets option | Workbook structure | Best document type | Post-export benefit |
|---|---|---|---|
| One sheet per page | Named tabs Page 1, Page 2 |
Mixed-layout reports, invoices | Keeps mismatched columns isolated |
| All pages on one sheet | Single tab named after file | Multi-page bank statements | Enables continuous SUM ranges |
When All pages on one sheet is selected, the tool inserts one blank separator row between consecutive pages so you can still spot page breaks before filtering. This single-sheet layout is ideal for annual tax summaries where you want one continuous SUM range across all twelve monthly statements.
If your PDF contains mostly narrative paragraphs rather than aligned columns, PDF to Word preserves headings, bullet lists, and flowing text far better than a grid.
Keeping bank statements and financial records offline
All page rendering, OCR recognition, and .xlsx ZIP compression (using Deflate level 6) execute locally inside your browser, sending 0 bytes over the network. Your sensitive bank statement is never uploaded to external servers, keeping payroll sheets, invoices, and tax records strictly on your device from drop to download.
Checking multi-line descriptions and OCR digits
Because PDF column detection aligns cells strictly by baseline and X-coordinate gaps, wrapped multi-line item descriptions or merged headers can split across two rows. Spot-check totals against the source statement, especially on scanned pages where OCR can confuse digits.
Tip: Open the exported
.xlsxin Excel or LibreOffice and run a quickSUMcheck on the amount column to confirm every row was captured and parsed as a number.
Working across Excel, Google Sheets, LibreOffice, and CSV
Unlike plain comma-separated files governed by the IETF RFC 4180 Common Format for CSV Files, .xlsx workbooks preserve separate sheets and explicit numeric cell types across Excel, Google Sheets, and Apple Numbers.
Choosing .xlsx prevents spreadsheet apps from misinterpreting dates or dropping leading decimals during import. If you later export a cleaned CSV from Excel and need a printable report, you can render clean CSV tables as PDFs locally without uploading your accounting data.