Two tools cover most jobs. Compress PDF shrinks a document by quality level, from High quality to Smallest, or aims for a target size in KB and lowers quality until it fits. It re-saves pages as pictures, so selectable text is lost; run OCR to PDF afterwards if you need searchable text. Compress image takes one photo or many and aims for the size you type in, for example 100 KB for a government form. Quality drops first, and the pixel size only shrinks if that is not enough.
When a form names an exact limit, the task pages are the quickest route. Each one opens the right tool already set up for 50 KB, 100 KB, 200 KB, 1 MB or whichever size you need, for PDFs and for JPG photos. If you are not sure what a portal accepts, read the guide on email and upload limits, which explains how caps work and how to get under them without ruining the document. The blog covers the rules of specific portals, such as USCIS, ImmiAccount and the UK visa upload.
Because the file is compressed by your browser, there is no waiting for an upload and no copy left on a server. It also means very large files depend on your device. If a result would not be smaller than the original, the tool keeps your original and says so.
Which route to pick: if the file must stay searchable, compress by quality level and then run OCR to PDF; if it only has to pass a limit, ask for a target size and let the tool find the quality. For a photo that ends up inside a PDF, compress the image first and then build the PDF with Images to PDF, so you control the size of each page picture. Open the result and scroll through it before you upload it, and check that signatures and small print are still readable.
A few habits help. Ask for the largest size a form allows rather than the smallest, because a larger target keeps more detail. Split a long scan or remove blank pages first if only part of it is needed. Everything on this page is free and needs no account; the only paid feature on DocPro is signing PDFs.