Converting and extracting

How to check a converted document before you send it

What should I check after converting a PDF?

Short answer

Check the things that break silently, not the things that break visibly. A missing page or a scrambled heading announces itself; a transposed digit in a total, a dropped minus sign, a date that came through as text instead of a date, and a table that gained a column do not. Five minutes spent on page count, figures, structure and the last page catches nearly everything that would otherwise be caught by the person you sent it to.

Converted files fail quietly

Every conversion from a PDF is an act of inference. The file records glyphs at coordinates; it does not record paragraphs, headings, columns, tables or reading order, so a converter works all of those out from geometry. Inference is usually right, and when it is wrong it is wrong in a way that still reads normally.

That is the problem. A sentence missing its second half looks obviously broken and gets fixed. An invoice total that reads 1,240.00 when the original said 1,249.00 looks completely fine, survives a careful proof-read, and is found by your client. The errors worth hunting are the ones that leave the document looking finished.

Start with the page count and the last page

Open the original and the conversion side by side and compare the page count first. It takes two seconds and it catches truncation, which is the single most common silent failure — a conversion that ran out of time partway through usually produces a perfectly valid file containing the first eighty pages of a hundred.

Then read the last page of both. Converters degrade towards the end rather than at the start, and the last page is where a file that was quietly cut short shows it. It is also where signatures, totals and appendices live, which is to say the parts somebody will act on.

If the counts disagree, check what the original actually contains before assuming the converter dropped something — printed page numbers and real sheets routinely disagree.

Check figures individually, never by reading

Numbers are where conversions do real damage, and reading through them does not work: the eye accepts any plausible digit. Check them deliberately.

  • Totals. Re-add one column. If the sum matches the original's stated total, the column came through.
  • Minus signs and brackets. Accounting negatives written as (1,240) frequently arrive as 1,240. The sign is a character like any other and it is the easiest one to lose.
  • Dates. In a spreadsheet, a date that landed as text will sit left-aligned and will not sort. Click one cell and look.
  • Long reference numbers. Account numbers, ISBNs and VAT numbers have no meaning to check against, so an error in one is invisible forever. Compare a few character by character.

If the document came from a scan, treat every digit as suspect: OCR confuses 0 with O, 1 with l and I, 5 with S, and 8 with B, and none of those substitutions makes a sentence look wrong.

Tables and headings, which is where the guessing happens

Tables are the hardest thing to recover from a PDF, because a PDF has no tables — only text and, sometimes, lines. A converter infers rows and columns from spacing, and gets it wrong in two directions: it finds a table in a block of prose, or it merges two columns into one because they were close together. Why PDF tables come out wrong in Excel covers this in detail; the check is to count the columns in the output against the original and look at any row where a cell is unexpectedly empty.

Headings are a gentler version of the same problem. They are detected by size, so a document that uses bold rather than a larger size for its headings converts into a wall of paragraphs. That is cosmetic, but it breaks the navigation pane and any table of contents built from styles — worth fixing before the document is handed to anyone who will navigate it.

Before you send it

Two last things, both about what the file carries rather than what it says.

Check the document properties. Converted files inherit the original's title, author and sometimes the name of software or a person who never touched your version. If the document is going outside your organisation, look at the properties before it goes.

Decide which file is the real one. A converted document is a derivative, and once two versions exist, somebody will edit the wrong one. Either the conversion becomes the master and the PDF is archived, or the PDF stays authoritative and the conversion is a working copy that gets deleted. Choosing is five seconds; not choosing costs an afternoon a few weeks later.

Tools this article covers

Sources

Primary documentation for the claims above.

Last reviewed: October 8, 2026

Published by iBuildPDF.

More from the knowledge base

Browse the knowledge base