Almost never. Reading, filling, signing, highlighting, reordering pages, combining and splitting are all built into what you already have. Buy something only for bulk work, text recognition at volume, or forms you are creating rather than filling. For another perspective on the same kind of decision, Monitask covers the topic in the full guide.
Reading and Marking Up a PDF
No affiliate links on this site, and nothing here is ranked.
What is already there
Reading, obviously, in the browser and in the operating system's own viewer.
Filling forms and signing. Both major systems do this without an account, and the result is a normal PDF anybody can open.
Highlighting and annotating, including on a phone or tablet, where a finger or a stylus is genuinely better than a mouse.
Reordering, rotating and deleting pages, which is where people most often assume they need a purchase.
Combining several PDFs into one, built in on one desktop system and available in the browser on both.
And exporting to an image or printing to a new PDF, which flattens annotations and is the way to send a marked-up copy that cannot be un-marked.
The three genuine cases
Bulk work. Splitting a hundred documents, extracting pages by rule, applying the same operation repeatedly. Manual tools do not scale and this is real.
Text recognition at volume. Making scanned documents searchable, across a library rather than one file. The phone does one page well; a thousand pages is a different task.
And creating forms — fields, validation, a document others will fill. Filling is free; authoring is the paid product, and the distinction is the one most people are unaware of when they buy.
What the paid tools also sell
Editing text inside a PDF, which works imperfectly on documents not produced for it and is frequently the wrong approach — the right one is editing the source and re-exporting, where the source exists.
Compression, useful and available free in the browser.
Redaction, which matters and where the free versions are dangerous: drawing a black box over text does not remove the text, and anybody can select it underneath. Genuine redaction is a case for a proper tool and for checking the result.
And conversion to editable formats, which produces variable results and is worth testing on your actual documents before relying on it.
The privacy point
Browser-based PDF tools upload your file to somebody's server unless they run locally in the page, which some do and few advertise clearly.
For a public document that does not matter. For a contract, a customer record or anything confidential it does — the same question as any tool holding your data, and it applies to a free web utility exactly as it does to a subscription.
Where the operation is available locally, use the local one. It is usually the same operation.
What to check if you do buy
Whether it does the specific thing you need, tested on your own documents rather than a sample. Scanned documents behave differently from generated ones, and a tool that handles one may struggle with the other.
Whether the output opens correctly everywhere. A PDF that only renders properly in the tool that made it is not a PDF for practical purposes.
And whether it is a subscription for something you do twice a year, which is the frequency question and which for this category usually argues against buying.
Sending a marked-up copy back
The step where the annotation is lost.
Annotations in some viewers are stored as a layer that another viewer may not show, or may show differently, or may allow the recipient to remove.
Print to PDF before sending, which flattens everything into the page and produces a file that looks the same to everybody.
Where the recipient needs to respond to specific comments, keep the layered version and say which viewer you used — or better, put the comments in the covering message with page numbers, which nobody can fail to see.
Scanned against generated
The distinction that explains most frustration with this format.
A generated PDF contains text. It is searchable, selectable and copyable, and tools work on it predictably.
A scanned PDF contains a picture of text. Nothing is searchable until recognition has been run, and every operation behaves differently.
Check which you have — try to select a word — before concluding a tool is inadequate. Most complaints about PDF tools are complaints about scanned documents.
The single most useful habit
Naming the file when you save it.
A folder of documents called scan001, invoice.pdf and document(3).pdf is a folder nobody can use, and the cost lands months later.
Supplier, subject, date. Ten seconds at the moment of saving, and it is the same rule as everywhere else on this site — the naming is the system, and no tool substitutes for it.
Where the format itself is the wrong choice
Worth saying on a page about handling it.
A PDF is a fixed page. It is right for anything that must look identical everywhere and that nobody should edit — quotes, invoices, signed agreements, certificates.
It is wrong for anything being worked on, and a document circulated as a PDF for comment produces a chain of annotated copies nobody can merge.
Send the editable format for anything collaborative and the PDF for anything final. Most of the difficulty in this category comes from using the fixed format for a moving document. For broader context, see Bitwarden.
The short version
- Reading, filling, signing, annotating, reordering, combining and splitting are all already installed
- Three genuine cases: bulk operations, text recognition at volume, and creating forms rather than filling them
- Filling is free and authoring is the paid product, which is the distinction most buyers are unaware of
- Editing text inside a PDF works imperfectly; editing the source and re-exporting is the better route where the source exists
- Drawing a black box over text does not remove it — genuine redaction is a case for a proper tool and for checking the result
- Browser-based tools usually upload the file, which matters for anything confidential and not at all for anything public