Do it in this order: fix rotation, remove blanks, merge, then compress once at the end. Every other sequence creates work you have to redo.
The mistake to avoid is compressing the parts before merging. It either re-encodes every image twice or leaves pages at uneven sharpness, where one pass over the merged document gives a consistent result.
You scan a thirty page contract on an office machine. What lands in the folder is eleven files, some with two pages and some with one, three of them sideways, six of them blank because the scanner ran in duplex mode and the paper was printed on one side. None of this is unusual, and none of it is your fault.
Turning that into a document someone can read is a sequence problem more than a tooling problem. Each step makes the next one cheaper, and doing them out of order means doing several of them twice.
The order that avoids rework
There is a reason this sequence works. Cleaning happens while pages are still in small files that are quick to open and easy to fix. Merging happens once the parts are correct. Compression happens last, when the tool can see the whole document and make one consistent decision about it.
Fix rotation first, file by file
A sideways page is easier to spot and fix in a three page file than buried at position forty-seven of a merged document. Rotate PDF shows a thumbnail of each page, so you can turn only the sideways ones before you combine.
Drop the blank backs
Duplex scanning of single-sided paper doubles your page count with empty sheets. Remove them now, while the file boundaries still tell you which scan they came from.
Merge into reading order
Load everything into Merge PDF, arrange the files deliberately, then combine. This is also the moment to notice a missing page, while the scanner is still next to you.
Compress the finished document once
One pass over the whole file beats several passes over parts of it, both for size and for visual consistency across pages.
Why the order comes out wrong
Scanner software names files by counter, timestamp or tray position, and none of those match reading order reliably. Then the merge tool sorts alphabetically and the result is scrambled in a way that looks random but is entirely predictable.
Text sorts one character at a time, so scan10.pdf lands before scan2.pdf. A batch that ran past nine files will always come out wrong unless you either pad the numbers or arrange the files by hand before merging.
01 through 11 takes less time than discovering the order is wrong after you have already compressed and sent the document.Blank pages that are not blank
Duplex scanners photograph both sides of every sheet whether or not anything is printed on the back. A thirty sheet document scanned in duplex becomes sixty pages, half of them empty.
The complication is that a scanned blank page is not blank. It is a photograph of white paper, which means it carries speckles, dust, the shadow of the paper edge and whatever bled through from the printed side. Automatic blank detection works on how much ink-like content a page has, and a slightly dirty scan can sit above the threshold while a lightly printed page sits below it. Removing blank pages from a PDF covers how to review them before anything is deleted.
Single-sided originals scanned in duplex produce one empty page for every printed one.
It is a picture of paper, so speckles and edge shadows still take up file size.
Dirty blanks can read as content, faint pages as empty. Always review before deleting.
Mixed page sizes in one document
Receipts, A4 letters and a folded plan scanned together produce a PDF where every page is a different physical size. It displays acceptably because readers scale each page to the window, and then it prints terribly because the printer does not.
This is worth fixing before the document goes anywhere near a printer or a filing system. If the pages are different sizes, putting them on one sheet size with the content scaled to fit gives a document that behaves predictably everywhere. Resize PDF does that for A4, Letter and four other sizes without re-rendering the scans.
Making the result searchable
A scan is a picture of text, not text. Search finds nothing in it, you cannot copy a line out of it, and a screen reader has nothing to read. That is true of every page in the merged document unless something adds a text layer.
Optical character recognition reads the shapes in the image and writes an invisible text layer behind the picture. The page looks exactly the same and becomes searchable. Do this after merging, so the whole document gets one consistent pass rather than a patchwork of separately processed parts. PDFslime's OCR tool runs Tesseract in the browser, so the scans are not uploaded for this step either.
- Text documents300 DPI, greyscale
- Documents with photos300 DPI, colour
- Archival reproduction600 DPI, colour
- Anything for email200 to 300 DPI
- Single-sided originalsSimplex, not duplex
- Output formatPDF, not JPEG
Before you send it
Scanned documents fail in ways that are obvious on screen and invisible in a file listing, so the last step is always to look at the thing you made.
Scroll the whole document fast, not page by page. Rotated pages and stray blanks are unmistakable at speed.
Check the page count. Thirty sheets should be thirty pages; sixty means the blanks are still there.
Search for a word you know is in there. Nothing found means the text layer did not happen.
Read the smallest text at 100 percent zoom. If the compression went too far, this is where it shows first.
Merge your scans in the browser
Combine as many files as you like, drag to reorder, and keep every page on your own machine.
Common questions
What goes wrong between the scanner and the finished file.
01What order should I do these steps in?
Rotate, remove blanks, merge, then compress once. Cleaning is cheapest while the files are still small and separate, and compression gives the best result when it runs once over the finished document rather than several times over its parts.
02Why did my scans merge in the wrong order?
Filename sorting. Text sorts character by character, so scan10 comes before scan2. PDFslime's Merge PDF has a Sort by name button that reads the numbers as numbers and puts scan2 first, but padding the names (scan02, scan10) fixes the order in every tool, not only that one.
03How do I get rid of the blank pages?
Identify their positions and extract everything else, which writes a new file without them. Review before deleting, because a scanned blank page carries speckles and edge shadows that can fool automatic detection in both directions.
04Why is my scanned PDF so large?
Because every page is a photograph. At 300 DPI an A4 sheet is 2480 by 3508 pixels, about 8.7 million. In colour that is about 26 MB of raw pixel data, and in greyscale about 8.7 MB. The scanner's own compression shrinks both a long way, and compressing the merged document handles the rest.
05Can I search the text in a merged scan?
Not until something adds a text layer. Optical character recognition reads the images and writes invisible text behind them, which makes the document searchable without changing how it looks. Run it after merging so the whole file gets one consistent pass.
06Should I compress each scan before combining them?
No. If you then compress the merged file as well, every image is re-encoded twice and the artifacts stack. If you do not, you are left with pages compressed at whatever settings each part got, which shows as uneven sharpness. Merge first, compress once.