Font and Page Logic in Office to PDF Conversion

When you convert a Word document to PDF the page count can change, tables can shift, fonts can look different. These aren't random errors; they're the natural consequences of moving between two different document models. This article explains those models and what changes during conversion, and why.

Two different document models

The flow document model: how Word works. The document is a stream of text and objects. The page layout is calculated at runtime: words are distributed into lines, lines into pages, as long as the flow continues. Add a word and everything reflows.

The fixed layout model: how PDF works. Every page is a fixed size and every element is placed at specific coordinates. The calculation is done once and the result is frozen.

Conversion is the move from the first model to the second: the flow is calculated once and the result is fixed.

The big advantage is that the document looks the same everywhere. The cost is that the calculation is made according to the conditions at conversion time.

Font metrics: the source of the shifting

How a piece of text breaks into lines depends on the widths of the letters. Every font carries its own metric information: how much space each character occupies, the spacing between letters, the line height.

When Word opens a document it uses those metrics to do the line-breaking math. If the font used in the document isn't on the system, Word picks the closest alternative — and the alternative's metrics differ.

The resulting chain:

1. Letter widths change.
2. Line break points shift.
3. Paragraphs occupy a different number of lines.
4. Page breaks shift.
5. The page count changes.

An extra page, a table split in two, or a heading stranded at the bottom of a page are all the result of that chain.

An example: in a 40-page report, if Arial is used instead of Calibri, Arial's wider letters can make the document 43 pages and every table breaks in a different place than before.

Font embedding

The solution to this problem is embedding: copying the font's glyph drawing data into the document.

PDF does this by default. It's the format's reason for existing — to guarantee the document looks the same everywhere, it carries the fonts used inside the file.

In Word embedding is optional and off by default because it increases file size. The "Embed fonts in the file" option lives under File > Options > Save.

There are two sub-options:

• Embed only the characters used: the file stays small, but you run into trouble if you edit the document and add new characters.
• Embed all characters: the file grows but you're completely safe.

License restriction: some commercial fonts don't permit embedding. The font file carries a permission flag and programs honor it. If you're using such a font, embedding fails and it falls back to an alternative.

Subset embedding in PDF

PDFs usually embed not the whole font but only the characters that appear in the document. That's called "subset embedding."

The advantage is file size: a full font family with broad language support can run to several megabytes, while a subset can be a few hundred kilobytes.

The side effect: when you later try to edit the PDF, you can't type a letter that doesn't appear in the document (if there's no "ğ" anywhere, say) — that glyph's drawing data isn't in the file.

Excel: the format with no concept of a page

Excel is the format that produces the most surprises, and the reason is structural: a worksheet has no natural concept of a page.

Theoretically it's a grid of unlimited width and height. The idea of a "page" only emerges through the print settings:

| Setting | Effect | |---|---| | Print area | Which cell range gets included | | Orientation | Portrait / landscape | | Scaling | A percentage, or "fit to N pages" | | Margins | The usable area | | Page breaks | Manually added break points | | Print titles | Rows/columns to repeat on every page |

PDF conversion uses these settings. If the settings aren't defined, Excel applies its defaults and the result usually isn't what you wanted.

Classic problems:

Hundreds of blank pages. If a space character was accidentally entered in a cell, Excel counts that cell as "used" and the print area extends all the way there. The result: dozens of blank pages. The fix: explicitly define the print area.

Cut off on the right. If the table doesn't fit the page width, Excel splits the columns and puts the rest on subsequent pages. The output is unreadable. The fix: "fit all columns on one page" scaling.

Unreadably small text. Excessive scaling makes text unreadable. Check the scaling percentage; values below 50% are usually a problem.

Missing headers. In a multi-page table the column headers disappear from the second page on. The fix: Page Layout > Print Titles.

The critical point: none of these problems are the converter's fault. The converter follows Excel's print settings. The fix has to be made in the source.

PowerPoint: losing the time axis

A presentation is a series of states changing over time: bullets appear in sequence, objects move, slides change with a transition effect.

PDF's content model has no time axis. Every page defines a single fixed appearance.

The converter draws the slide's final state — everything with all animations completed — as a single page.

The consequences:

| PowerPoint feature | In PDF | |---|---| | Object animation | The final state is shown | | Slide transition | Lost | | Embedded video | A static frame or an empty area | | Audio | Lost | | Interactive buttons | Usually lost | | Hyperlinks | Usually preserved | | Speaker notes | Requires a special layout |

The practical consequence: bullets that appear one by one through animation all show at once in the PDF. If you want step-by-step presentation, you need to prepare each step as its own slide — slide duplication instead of animation.

That's a common practice in presentation distribution: the presentation file is animated, while the PDF version is prepared with step-by-step slides.

What's preserved and what's lost

The structures that can and can't be carried over:

Preserved:

• Text (as real text, selectable and searchable)
• Vector graphics (shapes, lines, SmartArt)
• Images
• Tables (visually)
• Hyperlinks (in most converters)
• Internal links and the table of contents
• Bookmarks (can be produced from the heading structure)
• Page size and orientation
• Metadata (author, title)

Lost:

• Formulas (in Excel only the results remain)
• Animation and transitions
• Audio and video
• Macros
• The logic of tracked changes (can carry visually)
• Live data connections
• The logic of conditional formatting (the visual result remains)

May change:

• Page count (through font substitution)
• Image resolution (depending on compression settings)
• Color values (through color space conversion)

Tagged PDF and accessibility

Modern Office versions can produce a tagged PDF when generating the file. That adds the document's structural information — headings, paragraphs, lists, table cells, reading order — into the PDF.

The benefits:

• Screen readers read the document in the right order with the right structure.
• Text extraction and reflow become more reliable.
• Accessibility standards are met.

For public institutions and large companies that's usually an obligation. When saving as PDF in Word, make sure the "Document structure tags" option is on.

Using real heading styles in the source document (Heading 1, Heading 2) directly determines the quality of that structure. Just making text big and bold doesn't create a structural heading.

Image handling

Images in Office documents are affected by compression settings when transferred to PDF.

Word's "Compress images" setting can reduce images to a certain resolution when saving the document. The default is usually 220 DPI, which is enough for most uses; but if you're going to do high-quality printing, you need to turn that compression off.

The "Do not compress images in file" option lives under File > Options > Advanced > Image Size and Quality.

In summary

Office to PDF conversion is a move from a flow document model to a fixed page model: the page layout is calculated once and frozen. Because that calculation depends on the fonts available at conversion time, when a font isn't found the letter widths change and everything from line breaks to page count shifts — the fix is to use common fonts or embed them. In Excel the concept of a page exists only through the print settings, so the source of output problems is almost always the source file's settings. In PowerPoint the time axis is lost and animations are frozen in their final state. Preparing the source document with these limits in mind eliminates most post-conversion surprises.