Google is a multinational corporation that provides Internet-related products and services, including a particularly nice search engine, with US$ 50.18 billion in revenue. Google maps and Google local provide local information for millions of people everyday around the world.
Content originally written in English describing a variety of location based information changes over time in both English and in a number of languages that the content has been translated in.
Typically translation costs are calculated per word. Translators can’t easily tell what has changed in an article over time and charge the full article word count to translate the piece even if only a small piece has changed. At scale this simply is ineffective and the wrong thing to do.
There was a desire to calculate what should be translated and provide tools to both editorial and production staff to be able to budget and manage the workflow of translations. These tools would need to work across language boundaries and guide the translation process for the various stakeholders and translators.
We built a Drupal based solution that tracks versions of Articles and calculates the differences between versions in several languages. The editorial owners of the content can send new versions of articles for translation in particular languages and the system automatically picks the least booked translator that supports the language that the article is being translated into. It also lets the translator know how many words have been changed.
This slashes translation costs hugely as the system knows what needs translation and the translators only retranslated the relevant areas of the articles.
The client's view
Full Fat Things built “Transman” and amazed us with how fast something with complex business rules could be built to support our business workflow in Drupal. We worked with Full Fat Things on multiple projects but the Transman project was the big one that really changed the outlook on translations for us. - Grahame Oakland - Google
The initial engagement was stop and go which can limit product quality as developers push to deliver functionality and the mindspace required to hold all of the product up in the brain as you craft out something beautiful is harder to achieve when not attacking the problem constantly. As the project went on we learnt to agree specific fixed blocks of contiguous time to deliver functionality. This was the first time we used the real time project management tool Trello to power the project management. The real time nature of the product aided our communication and allowed us to spot opportunities for innovation with the client as we moved through the backlog of features we wanted to deliver.