Document usability
What is PDF reflow?
The short version: reflow is what should happen when a document stops behaving like a sheet of paper and starts behaving like text. The awkward part is that many PDFs were never built to do that well.
A PDF is usually a page, not a reading surface
Most PDFs are loyal to the page. That is their strength. A contract keeps its signatures in the right place. A research paper keeps its columns, figures, margins, footnotes, and page numbers. A restaurant menu looks like the printed menu.
Phones are not loyal to the page. They are narrow, held in one hand, used on trains, couches, sidewalks, and badly lit kitchens. When a full letter-size page is squeezed into that space, the text may technically be visible, but reading becomes work.
PDF reflow is the attempt to break that page loyalty when the reader needs text more than layout. It takes the text and document structure, then wraps it to the available width so you can scroll vertically instead of panning around a miniature print page.
Reflow is not magic resizing. It is a change in reading model: from fixed page to flexible text.
The accessibility world already has a name for the pain
WCAG calls this problem two-dimensional scrolling. The W3C explanation for reflow says the point is to let people enlarge text and related content without needing to scroll both sideways and vertically to read it.
That wording sounds dry until you try reading a dense PDF on a phone at 11 p.m. You zoom in. The line runs off the right edge. You drag sideways. You lose your place. You drag back. Repeat for the next line. The PDF is open, but the document is not really readable.
This is why reflow matters even for people who do not think of themselves as accessibility users. Low vision readers need it. Tired readers need it. Anyone reading a 40-page report on a small screen eventually understands the problem.
Good reflow depends on structure
A clean, tagged PDF has clues: headings, paragraphs, lists, tables, reading order, alternate text. A messy PDF may have only positioned text fragments. A scanned PDF may have no text at all until OCR runs.
This is the reason two PDFs that look similar can reflow very differently. One has a logical document underneath the visual page. The other is closer to a poster made from thousands of tiny text boxes.
A converter can infer a lot, especially with OCR and layout analysis. It still has to make choices. Should a sidebar interrupt the main text? Is this caption attached to the chart above it or the paragraph below it? Is the repeated line at the top of every page a heading or just a running header?
- Tagged PDFs usually give reflow more reliable reading order.
- Scanned PDFs need OCR before they can become flexible text.
- Tables, forms, diagrams, and legal layouts may need the original page view too.
Reflow is different from text extraction
Text extraction asks, What words are in this file? Reflow asks, What should the reader see next?
That difference matters. A raw extractor can return every word and still produce junk: broken line endings, column text stitched together, page numbers in the middle of sentences, captions floating away from images. The output is text, but not a reading experience.
A useful reflow pipeline is more editorial. It removes repeated page noise when it can, preserves headings when they are clear, rebuilds paragraphs, keeps lists as lists, and admits when a complex visual element should stay visual.
When conversion beats a viewer mode
Viewer modes are useful when you want to stay inside the original PDF app. Adobe Liquid Mode, for example, reformats headings, paragraphs, and images for mobile viewing without changing the underlying PDF.
Conversion is better when you want the reflowed result to become the thing you keep: a readable article, a Markdown export, a searchable library item, or a document you can revisit from another device.
The honest tradeoff: conversion may simplify design-heavy pages. For reading, that is often the point. For legal review, print proofing, or visual inspection, keep the original PDF beside the converted view.