Extract a table from PDF

Why this works

Our table extractor identifies tabular regions automatically, exports each table as Excel (.xlsx) or CSV, and stitches multi-page tables back together where rows continue across page breaks.

Table extraction is the surgical version of PDF-to-Excel. Where PDF to Excel converts every table in the document into separate worksheets, table extraction lets you target one specific table — the rate table in the middle of an insurance policy, the transaction grid on page 14 of a bank statement, the pricing matrix in a supplier quote PDF — and pull just that table without the surrounding noise.

Why that targeting matters. PDF documents often contain many tables that look similar but serve different roles. A 60-page annual report might have a summary table on page 3, ten detail tables in the financials section, and a quarterly comparison in the appendix. If you only need the quarterly comparison, running PDF to Excel produces a workbook with eleven sheets where you have to find the right one. Targeted table extraction lets you specify exactly which table you want, output as either a single-sheet .xlsx or a flat .csv ready for import.

Multi-page table stitching is the most under-appreciated feature. Long tables in PDFs frequently break across page boundaries — a transaction list that runs from page 14 to page 22 has the column headers repeated on each page and the data continuing where it left off. Most table extractors treat each page\'s table as a separate output, leaving you to manually concatenate them. Our extractor detects when adjacent pages share the same column structure and stitches them into one continuous output table, dropping the repeated headers automatically.

What extracts cleanly: simple grid tables with consistent column widths, financial statements with header rows clearly distinguished from data rows, transaction lists with stable column types per column, regulatory disclosure tables, price lists. Numbers extract as numbers (SUM-able in Excel), dates as dates (chronologically sortable), percentages as decimal equivalents (formula-ready), currency symbols as cell formatting.

Where extraction needs touch-up. Merged cells: detected and represented as Excel merged ranges, but complex merging schemes (header cells that span 3 columns over 2 sub-header rows) may need manual restructuring. Nested tables (a table inside a cell of another table): rare in PDFs, but when present they\'re flattened to plain text within the parent cell. Rotated text: if column headers are rotated 90° to fit in narrow columns, they may extract as horizontal text in the wrong position. Multi-table pages: when two tables appear side-by-side on the same page (e.g. a summary box alongside a detail table), the auto-detection sometimes treats them as one wide table with empty middle columns. Use the manual range adjustment to fix.

For scanned PDFs the workflow is two-step: run OCR first (using our OCR tool), then run table extraction on the OCR\'d output. OCR-then-extract gives noticeably better accuracy than relying on a single combined pass, because you can verify OCR text before depending on it for table structure.

Format choice between .xlsx and CSV. .xlsx keeps number formatting, cell colours from the source, and the multi-sheet layout if multiple tables export at once — the right pick for human analysis. CSV is plain text, one table per file, no formatting metadata — the right pick for piping into a script, a database import, or an automated processing pipeline.

How it works