On this page
Confirm the PDF contains selectable text
Try selecting a value in an ordinary PDF viewer. If the page is only an image, table detection has no characters or coordinates to group.
Use OCR PDF to Excel for scanned input, then review recognition errors separately from structural extraction errors.
Analyze only the pages that matter
A custom page range reduces noise in long reports and makes review faster. Use the active-page option while tuning detection sensitivity on a representative table.
Strict spacing avoids combining unrelated text. Relaxed grouping can recover wider columns but may create extra candidates.
Compare every candidate with its source region
The highlighted page region shows what produced the extracted grid. Check column boundaries, wrapped labels, negative values, footnotes and repeated headers.
A candidate is evidence to review, not a claim that business meaning has been understood.
Verify the exported structure
For XLSX, confirm the expected worksheet count. For CSV, confirm the delimiter and character encoding in the receiving application.
Reconcile totals and identifiers against the original PDF before using the data for analysis or decisions.
Put the workflow into practice
Choose a searchable PDF, tune the page scope, compare detected regions with editable grids, then export and verify only the reviewed tables.
Try Extract Tables from PDFFrequently asked questions
Clear answers before you extract tables
Why does a searchable PDF work better?
It exposes real positioned text items that can be grouped into rows and columns.
Can the detector understand merged cells?
Not reliably. Merged and nested structures often require manual cleanup.
Should I analyze every page?
Only when necessary. A focused range is faster and produces fewer irrelevant candidates.
Does extraction change the PDF?
No. It creates a separate data file and leaves the source unchanged.
