Safest: in-place text and image edits
Rewriting one text run or swapping the pixels of one image touches exactly that object. Everything else in the file, including pages you never opened, is written back as it arrived.
PDF reference
PDF layout preservation: a practical guide
Not every change to a PDF carries the same risk. Replacing a word is close to free; merging two documents is almost free; converting to another format and back is expensive every time. Most layout disasters are not caused by a bad tool but by reaching for a high-risk operation to accomplish a low-risk task.
Free — no signupNo watermarkNothing to installYour layout preserved
The short answer
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. Verify by comparing rendered pages before and after, not by reading them.
Safe: in-place edits and page operationsRisky: resizing, redaction, heavy compressionDestructive: conversion and rasterisationVerify by render diff, not by eyeCheck the page dimensions first
Why the risk differs so much by operation
This page ranks the operations you are likely to perform, says what each one actually touches, and ends with the four-step check that catches a broken layout before a recipient does.
Operations ranked by risk
Safe: page-level operations
Merging, extracting, reordering and deleting pages move whole pages around without touching what is drawn on them. The content of a page that survives the operation is unchanged.
Safe: encryption and metadata
Adding a password or changing document properties rewrites the file’s structure without altering any page’s drawing instructions.
Moderate: compression
Recompressing images changes how they look, not where anything sits. Aggressive settings soften photographs and can visibly degrade a scan, but the geometry of the page holds.
Moderate: redaction and resizing
True redaction removes covered objects, and on a page that is one large scan that means the whole page image. Resizing wraps the page in a transformation, which is safe for content but changes every physical dimension in the document.
Destructive: conversion and rasterisation
Round-tripping through another format re-derives your structure from scratch. Rasterising replaces text with pixels. Both preserve appearance at best and destroy everything underneath it.
The pre-send verification routine
Compare the page dimensions
Render the original and your export at the same scale. If the pixel dimensions differ, the page geometry was rewritten — stop there, because nothing else you check will matter.
Difference the renders
Subtract the two images. The change should be confined to the region you edited. A difference spread down the page means content moved.
Copy the text you changed
Select the edited line in the exported PDF and paste it somewhere plain. Your new wording proves a real edit; the old wording proves an overlay was drawn on top of text that is still in the file.
Copy text you did not change
If text elsewhere now extracts differently, the whole document was re-encoded on save — which affects pages you never opened.
Open it somewhere else
A phone, another browser, a colleague’s machine. This is the only test that catches fonts your own computer happens to have installed.
The risk table
What each common operation actually touches. Risk here means risk to the layout specifically, not to the document overall.
| What it touches | Layout risk | |
|---|---|---|
| Edit a text run in place | One text object | Minimal |
| Replace an image in place | One image object | Minimal |
| Merge, split, reorder, delete pages | The page tree only | Minimal |
| Annotate, highlight, sign | Adds new objects | Minimal — nothing is removed |
| Encrypt or edit metadata | File structure, not pages | Minimal |
| Compress | Image data | Moderate — quality, not position |
| Redact | Removes covered objects | Moderate — can drop a whole scan |
| Resize pages | Wraps content in a transform | Moderate — every dimension changes |
| Convert to Word and back | Rebuilds the entire document | High |
| Rasterise pages | Replaces text with pixels | Total — appearance only survives |
Where preservation is impossible
Scanned pages have no layout to preserve
A scan is one image. There is no text, no structure and nothing to protect beyond the pixels themselves — which is also why nothing can shift on one.
Substantial rewrites need the source file
If a paragraph has to double in length, or a section has to move, the layout has to change. Preservation is the wrong goal for that job; go back to the document the PDF was made from.
A perfect render diff is not a perfect document
Pixel identity proves nothing moved. It says nothing about whether the text is extractable, the fonts are embedded, or the tagging survived. Those are separate checks, and a few of them are free tools on this site.
Habits that prevent most of it
Pick the narrowest operation that does the job
Almost every horror story on this subject is a wide operation used for a narrow task. If you need to change a word, change the word.
Keep the shipped PDF as the working copy
Editing the file that actually went out guarantees the corrected version differs in exactly one place. Rebuilding from source introduces differences nobody asked for.
Verify before you send, not after
The render diff takes a minute. Recalling a document does not.
Embed fonts at export time
It costs kilobytes and removes the entire class of problems where a document is fine on your machine and wrong on everybody else’s.
Frequently asked questions
Anything that leaves existing objects in place. Editing a text run, replacing an image, merging or splitting documents, reordering or deleting pages, annotating, signing, and adding encryption all touch a narrow part of the file and leave every page’s drawing instructions alone. The pages that survive the operation come out exactly as they went in.
Anything that rebuilds the page. Converting to Word, Google Docs or another format and back re-derives structure the PDF never stored. Rasterising replaces text with pixels. Both are high risk by construction, not because of any particular tool’s quality.
Render the original and the exported file at the same scale and compare the images. First check that the pixel dimensions match — different dimensions mean the page geometry was rewritten. Then difference the two: the change should be confined to the area you edited. A difference spread down the page means content moved.
Yes for geometry, with a caveat for quality. Compression works on image data, so nothing moves, but aggressive settings soften photographs and can visibly degrade a scanned page. The layout is preserved; the fidelity of the images may not be.
It can, in one specific way. True redaction removes the covered content rather than hiding it, and because a raster image cannot be partially cropped, redacting into a full-page scan removes that page image entirely. On a normal text page it removes only the covered text runs and leaves everything else in place.
Not in the sense that content shifts relative to the page — a resize wraps the existing content stream in a single transformation, so text stays selectable and proportions hold. But every physical dimension in the document changes, which matters for print, for signing portals and for anything that expects a specific page size.
Converting to Word to make a small change. It is the highest-risk operation available applied to the lowest-stakes task, and it accounts for most of the collapsed tables, shifted columns and substituted fonts people encounter.
Perfectly, for the right operations — and that is measurable rather than a matter of opinion. Editing one line of a sample invoice changes 0.02% of the rendered page, with everything below the edit pixel-identical. The benchmark page linked below shows the method and the images.
Tools and guides that go deeper
Other editing problems, solved
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.
PDF editing fidelity: a measured benchmark
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.
Why PDF editors break formatting
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.
Edit a PDF without losing its formatting
To edit a PDF without losing formatting, edit the PDF itself instead of converting it to Word and back.
Replacing PDF images without breaking the layout
A PDF draws an image by stretching it into a rectangle defined by a transformation matrix, so a replacement inherits the original’s frame regardless of its own proportions.
Ready when you are
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