PDF guides
How PDF to Excel Detects Tables
PDF to Excel does not know in advance which parts of a page are a "table" — a PDF has no built-in concept of rows and columns, only text positioned at coordinates. This guide explains, at a practical level, how the tool infers table structure from that raw positioning, how it decides numbers from text, and when a table is judged to continue across a page break.
Ready to try the tool this guide describes?
From positioned text to a grid
Text on each page is first grouped into visual lines by vertical position. The tool then looks for runs of consecutive lines that share the same left margin and have recurring gaps at the same horizontal positions across most of those lines — a strong signal that the page contains a real grid rather than ordinary paragraph text.
Why plain paragraphs are not mistaken for tables
A justified paragraph can share a left margin across every line, but it does not have recurring gaps in the same interior positions from line to line — its wrapping is driven by word length, not columns. The detector specifically looks for that repeated interior alignment, so ordinary prose is not misclassified as a table.
How cell values are typed
- Plain numbers, grouped-thousands numbers, and unambiguous currency amounts (like $1,234.56) become real numeric cells.
- Unambiguous dates — ISO dates like 2024-01-05, or dates with a written month name — become real date cells.
- Percentages, leading-zero codes (like ZIP codes), long ID or account numbers, and locale-ambiguous slash-formatted dates are kept as text on purpose, since converting them could silently produce a wrong or misleading value.
Merging a table across a page break
When a table appears to continue from the bottom of one page to the top of the next, the tool only merges the two into a single sheet if the column count matches and the column positions clearly line up between the two pages. If that evidence is not clear, the two are kept as separate tables rather than risking a merge of unrelated data.
PDF to Excel Table Detection FAQ
- How does the tool know what counts as a table?
- It looks for consecutive lines of text that share a left margin and have gaps recurring at the same horizontal positions across most of those lines — a pattern normal paragraph text does not produce.
- Why is a percentage kept as text instead of a number?
- Converting "15%" to a number is ambiguous — it could mean 15 or 0.15 depending on context. To avoid guessing wrong, percentages are kept as their original text.
- Why are ZIP codes and ID numbers kept as text?
- Converting a value like "00501" to the number 501 would silently lose the leading zero. Leading-zero numeric strings and long ID/account numbers are kept as text to preserve their exact original form.
- Why are some slash-formatted dates not converted?
- A date like "3/4/2024" is genuinely ambiguous between month/day and day/month ordering depending on locale. Only unambiguous formats — ISO dates and dates with a written month name — are converted to real date values.
- Will a table split across two pages always merge into one sheet?
- Only when the column count and column positions clearly match between the two pages. Otherwise the tool keeps them as two separate tables rather than risk combining unrelated data.
Related guides
- How to Convert PDF to ExcelPull tables and data from a PDF into an editable .xlsx workbook in your browser. Covers how-to steps, what gets reconstructed, and key limits.
- PDF to Excel Scanned PDF LimitationsWhy scanned or image-only PDF pages have no extractable table data, how PDF to Excel detects and handles mixed documents, and why this tool does not perform OCR.
- PDF to Word Table ReconstructionHow PDF to Word detects simple, regularly-gridded tables and rebuilds them as real Word tables on a best-effort basis, and when a table comes through as plain text instead.