Converting Word, Excel and PowerPoint to PDF

Going to PDF

This direction is the easy one. Your application has already computed the layout in order to show it to you, and producing a PDF records that result along with the fonts. What you see is what the file contains.

• Word to PDF — documents, letters, reports.
• Excel to PDF — spreadsheets, with the caveat below.
• PowerPoint to PDF — slides, one per page.

Two things still go wrong. Unusual fonts may be substituted if they cannot be embedded, which changes line breaks — the same failure described in why PDFs look different on other computers. And the result can be much larger than the original, because Office formats compress images hard; Compress PDF brings it back down.

Coming back, and what it costs

Going the other way, the converter is handed marks at coordinates and has to rebuild meaning: which glyphs formed a word, which lines formed a paragraph, whether aligned text is a table or a coincidence.

It infers, and inference varies with how the file was made. A PDF produced by Word converts back well, because its structure was regular. One produced by a design application, or a scanner, converts back badly.

Which means:

• Keep the original file. Every round trip you avoid is a round trip that cannot go wrong.
• Expect to fix the layout after PDF to Word. What a converter can and cannot recover sets expectations properly.
• If the PDF is a scan, convert nothing until you have run OCR. There is no text in the file, so a converter will hand you a document containing photographs — correctly, and uselessly.

Spreadsheets are a special case

A spreadsheet is not really a document. It is a grid with formulas, and printing it means deciding where to cut a grid that has no natural page boundaries. That decision is why Excel to PDF so often produces a file with columns stranded on their own pages.

Set the print area, scaling and orientation in the spreadsheet before converting. Those settings are what the conversion follows, and no amount of adjustment afterwards recovers a column that was cut off.

Formulas do not survive. A PDF records the values that were showing, not the calculations behind them. Converting back with PDF to Excel recovers a grid of numbers and text, and every formula is gone — the PDF never contained them. For the same reason, a PDF whose tables are drawn with lines rather than laid out as a grid may come back as text rather than as cells.

Slides and handouts

PowerPoint to PDF gives one page per slide, which is what you want for distributing a deck: it opens anywhere, it cannot be accidentally edited, and the fonts travel with it.

What does not travel: animations, transitions, embedded video and audio. A build that reveals four bullets one at a time becomes one page with four bullets. Speaker notes are usually dropped unless you explicitly export a notes layout.

Coming back, PDF to PowerPoint produces editable slides, but each one is reconstructed — text boxes land approximately, and the original layout masters are gone. It is a reasonable way to salvage a deck you no longer have the source for, and a poor substitute for keeping the source.

If you want several slides per sheet for printing, export handouts from the presentation software rather than converting and then trying to rearrange pages. For combining decks afterwards, Merge PDF is the tool.