A scanned PDF can often fit under 2 MB while keeping the words easy to read. Start with the best originals you have and preserve a copy before you edit. Use the upload instructions to check whether 2 MB is the limit for one file or the whole batch. If you are close to the cap, aim for a little headroom so the file is accepted even when the operating system and website display sizes differently.
Find the pages that need attention
Skim every page at the same zoom. A color brochure, a receipt on patterned paper and a clean typed page have different storage needs. One full page color photograph can occupy most of a file that otherwise contains simple text. If a PDF editor can show individual page sizes or image sizes, use that information to identify what to change first. You can also compare exports of individual pages in a temporary working folder. Keep those temporary copies separate from the final PDF so you do not submit the wrong file.
Rescan for clean edges and even light
When the paper original is available, make a fresh scan instead of starting with a blurry copy. Flatten the paper and hold the phone parallel to the surface. Use broad, even lighting and move the light if the paper shows glare. Place pale sheets on a dark surface so the scanner can identify the edges. Keep a small margin around the page instead of letting a tight crop cut off footnotes or serial numbers. Correct orientation before combining the pages. These steps improve legibility before compression starts.
Choose resolution and color by page
Try scanning ordinary printed text at about 300 dots per inch as a starting point. Inspect small lettering before reducing resolution further. Use grayscale for pages where the original ink colors do not matter, and keep color for diagrams, stamps or handwritten notes whose colors convey meaning. A crisp monochrome scan can be compact, but a hard threshold can erase faint marks. Test one representative page first, then compare it with your source at an identical zoom. Treat pages with small type or colored labels independently if your software allows it.
Keep searchable text when possible
Optical character recognition can make a scan searchable by adding a text layer. It cannot make blurry letters visible to a reader. If a scanned PDF was already searchable, check that the compressed copy still lets you find a phrase. Read that phrase in the image, too. If your compression tool turns each page into a new flat image, it may discard the text layer. Use a different export option when searchability matters. For a PDF that already contains live text, optimize embedded images rather than rasterizing the whole document.
Work toward a size budget
Suppose you need to submit eight pages in one PDF. A 2 MB limit provides roughly 250 KB per page on average before accounting for the document overhead. A simple page can use less of that budget, leaving space for a detailed table or a signature. Start by reducing the largest photo pages. Try one adjustment at a time, such as changing scan color or lowering image quality one step. Compare the resulting size with the previous export. If the text begins to show broken strokes or fuzzy edges, return to the original and try a different adjustment.
Review the PDF on a second screen
Open the exported copy at a normal reading size, then zoom into the most difficult page. Check small numbers, punctuation, table lines, initials and faint handwriting. Confirm that no page is upside down, repeated or missing. If the document has an electronic signature or fillable fields, verify those details before replacing anything. Do not upload the edited version as a signed original unless the recipient accepts it. Keep the source in a separate folder with a clear filename.
Attach and confirm
Check the exported PDF’s exact size in your file manager. Use a filename you can distinguish from earlier copies and pick that file in the upload form. If the portal provides a preview, open the first and last pages and verify the order. If a file that fits still fails, look for restrictions on passwords, page counts or total attachment sizes. If the only version under the cap is unreadable, keep the readable source and check whether the receiving portal offers another submission option.