Merge PDFs or select pages: a checklist for the final document
Combine PDF files, extract required pages and correct orientation while checking order, page ranges, signatures and upload limits.
The form says "upload supporting documents" and offers exactly one file input. You have four PDFs: a bank statement, a lease, a tax notice, and a two-page appendix the instructions happen to mention.
Or the opposite situation: someone sends you a 90-page scan and asks for "pages 12 to 18" — and you realize the printed page numbers stopped matching the actual PDF pages somewhere around the third section.
Merging and page selection solve the two halves of this problem. Neither one makes the document smaller, and knowing that in advance saves a lot of wasted effort.
Merging: order is the whole job
Merging concatenates documents. If the inputs contain 2, 3, and 1 pages, the output contains 6. Nothing is re-encoded, nothing is optimized, and the file will be roughly the sum of its parts.
That arithmetic is your cheapest test. Write down the page count of every input before you start, add them up, and compare against what you get. A missing page almost always means a file you forgot to select, and you will catch it in five seconds instead of after a rejection.
Controlling the order
The order you want is rarely the order your file dialog hands over. It depends on the dialog, the operating system, and the sequence in which you clicked. Treat the initial order as a draft, not a result.
The reliable approach is to rename working copies with numeric prefixes — 01_bank_statement.pdf, 02_lease.pdf, 03_tax_notice.pdf, 04_appendix.pdf — select them, then confirm the order in the workspace before you run anything. Sorting by filename in your file manager also makes the selection order more predictable. Then verify again in the output.
If the order comes out wrong, do not re-run and hope. Fix the names, rebuild, and check again. Merging is cheap; a submission with the lease before the cover sheet is not.
What happens at the joins
Page sizes can differ between documents, and merging does not reconcile them. If the bank statement is A4 and the appendix is Letter, the merged PDF has both, page by page. Most readers cope with that without complaint.
Portals are less forgiving. Plenty of upload validators check that all pages share the same dimensions, and some check that they match a specific standard. If your target document will be validated that way, check each source's page size first with PDF inspection, and decide whether to accept mixed sizes or rebuild the odd document at the right size.
The other join problem is content. When two documents meet, the seam is often invisible — page 4 ends one report and page 5 begins another with no marker. If the recipient needs to see where one document ends, consider whether a separator page is required, and whether the instructions say anything about it.
Selecting pages without trusting the printed numbers
The page selection tool works on PDF page positions, counted from 1. It has no idea what number is printed in the corner of the page.
That distinction bites everyone. A report with a cover page, a contents page, and a section divider has three unnumbered pages before page "1" appears — so printed page 3 is PDF page 6. In a scan where every page is an image, there may be no printed numbers at all.
So find the pages visually first. Open the document in a normal reader, scroll to the content you need, and note the actual position. Then enter the range.
Extracting, reordering, and repeating pages
The page expression is the whole interface, and its order is the output order. The syntax follows a simple comma-separated pattern:
| Expression | Result |
|---|---|
1-3,5 | Pages 1, 2, 3 and 5 — four pages |
3,1,2 | The same three pages, reordered |
7,7 | Page 7 twice |
That last one is more useful than it sounds. If a form wants a signature page at both the front and the back of a submission, repeating the page is a legitimate way to do it — you are duplicating an existing page that already contains the signature image, not re-signing anything.
Two rules to keep in mind. Reversed ranges are rejected, so 5-3 is not accepted; list the pages you want in the order you want them. And repeated pages count against the output page limit, so 7,7,7,7 costs four.
Why splitting will not shrink the file much
Here is the part that surprises people: extracting 10 pages out of a 90-page PDF often produces something much larger than a tenth of the original.
The reason is that pages hold images, and images carry most of the weight. If your 90-page scan is 40 MB and relatively evenly spread, keeping 10 pages might give you 5 MB — proportional. But scans are rarely even. A few high-resolution pages, like a stamped certificate or a colour photograph, can account for most of the bytes. Extract only those and you have barely saved anything, because you kept exactly the heavy pages.
There is also per-page overhead that does not divide neatly: fonts, embedded resources, and document-level structures may be partly rewritten when pages are removed. The output is a fresh document, not a sliced copy.
So use page selection for what it is good at — removing pages the recipient did not ask for, reordering, isolating one part of a long scan — and use PDF compression when the goal is actually the byte count. If the destination does not need the omitted pages, deleting them is still worthwhile: fewer pages is less data, even if the saving is not proportional.
A quick way to find out
Inspect the document before you select pages. PDF inspection reports page count, page dimensions and the embedded images, which tells you where the weight sits. If three pages hold most of the megabytes, selecting the other eighty will not rescue your file size, and you should plan for compression instead.
When rotation is the actual fix
A sideways page is a common artefact of scanning and phone photography, and PDF rotation handles it: 90, 180 or 270 degrees, applied to the whole document or a specific range.
Two things it does not do. It does not improve clarity — a blurry scan stays blurry, and rotating a skewed page leaves it skewed. And it changes display orientation, not content, so text and geometry are preserved rather than re-rendered.
Apply rotation after you have the page set you want, and then look at the result. Documents that mix landscape tables with portrait text often rotate better in one direction for some pages and the other direction for others, and it is worth handling those separately rather than applying one rotation to everything.
The final check
Before you upload, walk through this list:
- The first page and the last page are the ones you expect.
- Every required document is present, in the intended order, and the joins are where you think they are.
- Page dimensions and orientation satisfy the receiving site's stated rules.
- The byte count is under the limit.
- Anything with a digital signature has been handled through the path the recipient accepts.
That last point deserves care. Rewriting a signed PDF can invalidate its signature, and merging, page selection and compression all involve rewriting. Never overwrite your only signed original — work from a copy and keep the signed version intact. If the recipient requires a valid digital signature, the only safe process is the one they published.
One constraint to plan around: AttachReady rejects encrypted and password-protected PDFs, plus documents containing active content such as embedded scripts or attachments. If a source file is locked, export a clean copy from the application that produced it — password removal is not something these tools do.
And when you are working toward a byte target, remember what the target means. The 200 KB preset is 200,000 bytes, and current guest and account ceilings are listed alongside it. Related reading: compression boundaries for scanned PDFs.