Word to PDF — faithful conversion without Microsoft Office on your server

What it does

The API takes a .docx file and returns a PDF with layout-faithful conversion. No MS Word and no LibreOffice daemon of your own needed — the rendering engine runs server-side, deterministically and horizontally scalable.

Bonus: placeholders can be substituted before conversion — so a contract template gets filled automatically with customer data.

Minimal request

curl -X POST https://api.dokmatiq.com/v1/convert/word-to-pdf \
  -H "Authorization: Bearer $DOKMATIQ_KEY" \
  -F "document=@contract-template.docx" \
  -o contract.pdf

Response: Content-Type: application/pdf, status 200.

With placeholder substitution

A .docx with placeholders like {{customerName}}, {{date}}, {{amount}}:

curl -X POST https://api.dokmatiq.com/v1/convert/word-to-pdf \
  -H "Authorization: Bearer $DOKMATIQ_KEY" \
  -F "document=@contract-template.docx" \
  -F 'variables={"customerName":"Acme AG","date":"2026-04-18","amount":"€12,500.00"}' \
  -F "outputProfile=PDF/A-3b" \
  -o contract-acme.pdf

The placeholder syntax works with text, but also with images ({{logo}}), tables (row-wise with {#items}...{/items}) and simple conditions ({#isPremium}...{/isPremium}).

What converts cleanly

• Fonts (embedded or replaced with metric-compatible ones)
• Headers and footers, page numbers
• Tables of contents (with page references, clickable inside the PDF)
• Tables with cell merges, borders, shading
• Embedded images (PNG, JPEG, SVG)
• Hyperlinks
• Footnotes, endnotes
• Comments (optionally hidden)
• Tracked changes (optionally accepted or kept as annotations)

What needs extra attention

• Exotic OLE objects (embedded Excel charts): converted as static images
• Macros (.docm): ignored (security), content is preserved
• Password-protected sections: must be unlocked
• Non-standard fonts: only loaded server-side from a curated font pool; missing fonts fall back through a substitution table

What it’s used for

• Contracts from Word templates with CRM data
• Quotes, orders, invoices from SAP/DATEV exports
• Reports, appraisals with long prose and figures
• Government letters, legal notices built from legal text modules
• Mail merge — one template, hundreds of personalisations

Comparison with alternatives

	Dokmatiq Word-to-PDF	Microsoft Office (server)	LibreOffice headless	docx2pdf (Python)
License	included	commercial, restrictive	LGPL	MIT, needs Office
Scaling	API-side	hard on Linux	daemon management	local, not parallel
Layout fidelity	high	highest	good, occasional quirks	only with Office
Placeholders	built-in	via OpenXML SDK	external	external
Deterministic	yes	no (auto-update risk)	mostly	mostly
PDF/A	flag-driven	limited	yes	no

Running Microsoft Office server-side is legally murky and technically brittle. A LibreOffice daemon is a valid alternative but quickly turns into an operations concern (crashes, font management, zombie processes).

Common pitfalls

1. TOC with wrong page numbers — run field updates in Word before upload; otherwise the PDF shows the old page numbers
2. Missing corporate font — register corporate fonts via the font-upload endpoint; otherwise the fallback kicks in
3. Track Changes accepted or rejected? — default is “accepted”; for review scenarios use trackChanges: "visualize" to render as annotations
4. Encoding issues — .docx is always UTF-8; hand-crafted OOXML files sometimes violate that and garble umlauts