Case Study: Polhus achieves up to 75% publication-ready AI translations and $80K savings with Crowdin

Achieving Up to 75% of Translations Ready for Publication with AI – Polhus Localization with Crowdin

Polhus, a prefabricated wooden houses company, faced a major localization challenge after redesigning its website and overhauling Swedish content: retranslate roughly 1.6 million words across seven languages and 30+ CMS data models while preserving technical terminology and quality. To meet this need they partnered with Crowdin, using its AI-powered localization workflow and the Crowdin app for DatoCMS to synchronize and manage content for translation.

Crowdin implemented a glossary-first approach, integrated directly with DatoCMS, and used AI pre-translation (OpenAI ChatGPT 4o) followed by human review from eight expert linguists. The result: up to 75% of translations were publication-ready without edits, 100% were reviewed, turnaround measured in days, about $80,000 saved and dozens of hours conserved, with all 1.6M words synced back to the CMS—delivering fast, consistent, high-quality localization for Polhus.

Polhus

Anton Troedson

Chief Technology Officer