PDF reference
"It broke my formatting" describes a symptom, not a cause, which is why the usual advice — try a different tool — works about half the time. There are five distinct mechanisms, they produce different symptoms, and once you can tell them apart you can predict which tools will fail on which documents.
PDF editors break formatting through five distinct mechanisms: converting the file to another format and back, re-saving the whole document rather than appending a change, re-flowing the edited line from scratch, substituting a font the file does not contain, and rasterising pages into images. Each produces a different, recognisable symptom.
Underneath all five is one fact about the format. A PDF describes a page as a set of drawing instructions at fixed coordinates. It contains no paragraphs, no columns, no tables and no concept of text flowing around anything. That rigidity is why a PDF looks the same everywhere — and it is why any operation that has to reconstruct structure the file never stored is a guess that can go wrong.
PDF to Word, edit, Word to PDF. Each conversion re-derives structure the file never stored — where columns start, which lines form a table, which font is close enough. Two guesses for one word change, and the most destructive cause by a wide margin.
Many editors rebuild and rewrite the entire document on save rather than appending the change. Objects get renumbered and re-encoded, and things that were fine can shift. An incremental append leaves the untouched parts of the original bytes exactly as they were.
A PDF positions runs of glyphs at fixed coordinates, sometimes with per-character adjustments. An editor that rebuilds the line has to re-derive every one of those positions. Small errors show up as words bunching together, gaps opening mid-sentence, or the line drifting off its baseline.
The document embeds only the characters it used. Type one it never contained and something else has to draw it. A substitute with different widths does not just look different — it shifts every glyph after it on the line.
The nuclear option some tools take: render each page to an image and wrap it back in a PDF. Nothing can move, because nothing is text any more. Search, copy, accessibility and file size all pay for it.
Editing the existing text object in place, reusing the document’s own embedded font, writing at the original coordinates, and appending rather than rewriting. None of the five mechanisms gets a chance to fire.
If content below your edit shifted, the page was re-flowed or re-built. If only the edited line looks wrong, the cause is local — a font substitution or a re-flowed line.
If the text on the exported page cannot be selected at all, the document was rasterised. This is unmistakable and unrecoverable without OCR.
Render the original and the export at the same scale. Different dimensions mean the page geometry itself was rewritten, which points at a full re-save rather than a targeted edit.
Different stroke weight or letter widths where your edit meets untouched text means two font programs are drawing one line — cause four.
If text elsewhere in the document now copies out differently from before, the file was re-encoded on save. That is cause two, and it affects pages you never opened.
Match the symptom to the mechanism — they do not overlap much.
| Symptom you see | Cause | |
|---|---|---|
| Tables collapsed or cells merged | Structure was inferred and got it wrong | 1. Conversion round trip |
| Content below the edit moved down | The page was laid out again | 1 or 2. Conversion or full re-save |
| Words bunched or gaps mid-line | Glyph positions were recalculated | 3. Line re-flow |
| One line in a different typeface | A glyph was missing from the subset | 4. Font substitution |
| Text can no longer be selected | The page is now an image | 5. Rasterisation |
| File size jumped several times over | Text instructions became pixels | 5. Rasterisation |
| Pages you never opened changed | The whole file was rewritten | 2. Full re-serialisation |
A PDF-to-Word converter is asked to reconstruct a structure that was thrown away when the PDF was made. It gets it right often enough to be tempting and wrong often enough to cost you an afternoon. The mistake is using conversion for a job that does not need it.
Scanned pages have no text to edit. A few internal structures cannot be written back to safely. The right response to those is a clear message, not a best effort that produces a damaged file.
If you replace three words with fifteen, the line has more content than it was laid out for. That is not the editor breaking your formatting — it is the request requiring more space than exists.
Correcting words in a finished document is an editing job. Restructuring a document is a source-file job. Most formatting disasters come from using the second approach for the first task.
It is the single highest-risk operation on this page, applied to the lowest-stakes change. If the layout is already right, do not put it through two format translations.
Editing the PDF that actually shipped guarantees the corrected version differs in one place. Rebuilding from source introduces differences nobody asked for.
Rasterise the original and your export at the same scale and compare. It takes a minute and catches all five mechanisms at once.
Because a PDF stores a page as drawing instructions at fixed coordinates, with no record of paragraphs, columns or tables. Any operation that has to reconstruct that structure — most commonly converting the file to Word and back — is making educated guesses, and the formatting you lose is the difference between the original coordinates and the ones the guess produced.
Because nothing in the file says "this is a table". A table in a PDF is a set of text runs and ruled lines that happen to be positioned in a grid. A converter has to infer the grid from those positions, and when the inference is slightly off, cells merge, columns shift or rows split. Editing text in place never has to make that inference at all.
Because the line was rebuilt. PDFs position runs of glyphs individually, sometimes with per-character kerning adjustments, and an editor that redraws the whole line has to re-derive every position. Small errors in that calculation surface as words bunching up or unexpected gaps. Writing the replacement at the original line’s coordinates avoids the recalculation entirely.
The pages were rasterised — rendered to images and wrapped back in a PDF. Some tools do this because an image cannot reflow, so layout is trivially preserved. The price is that search, copy, accessibility and file size are all destroyed, and getting text back requires OCR.
Because the tool re-serialised the whole document on save rather than appending your change. Rewriting the file renumbers and re-encodes objects across every page, so content you never touched can be subtly altered. An incremental save leaves the original bytes of untouched pages exactly as they were.
Edit the existing text object in place: keep the document’s own embedded font, write at the original line’s coordinates, append the change rather than rebuilding the file, and never convert or rasterise. That combination is what this editor does, and it is measurable — see the fidelity benchmark linked below.
Any tool can, because the mechanisms above are properties of the operation rather than the vendor. Editing in place is low risk in any application that does it; converting to another format and back is high risk in any application that does that. What matters is which operation you are actually performing.
Check three things. Is there selectable text, or is the page a scan? Are the fonts embedded, and are they subsets? Does the change you need fit roughly in the space the old text occupied? The free font and layout checkers on this site answer the first two in a few seconds.
Every one of these opens the same free online PDF editor — each page covers what makes that particular problem hard, and what to do about it.
Editing one line of a sample invoice changes 0.02% of the rendered page: 399 pixels out of 2,009,712, confined to a 130 by 13 pixel region.
To edit a PDF without losing formatting, edit the PDF itself instead of converting it to Word and back.
PDF layout survives operations that leave existing objects in place — editing a text run, replacing an image, merging, extracting or deleting pages — and breaks under operations that rebuild the page: converting to another format, rasterising, or re-flowing text.
PDF font substitution is a viewer or editor replacing an unavailable font with a different one so the text can still be drawn.
Free, no signup, no watermark, nothing to install — and no round trip through another file format to get your document back.
Open the PDF editor