Boosting sales through the use of MT plus
Founded in 1994 as an online bookstore, Amazon is now a world-renowned household brand with an international presence. During its first month of trading the business fulfilled orders in 50 US states and 45 countries. Two decades later, by virtue of their mission to be a customer-focused company offering convenience, variety and competitive pricing, they are now the world’s biggest online retailer, with offices in 12 countries and over 200 million active customers. Technological innovation, which has been central to their expansion, has led to many additional offerings such as web services, credit cards and the well-known Kindle.
Amazon has seen a particularly swift expansion in Europe over the last decade, and in order to maintain this rate of growth they need to continually update and maintain their product catalogue. Many products are offered in one country on first launch, before then being made available to customers in other markets as well. Not only does this require time, personnel and infrastructure, it also requires Amazon to put in place a flexible and cost-effective method for translating the associated product descriptions and features.
During the key selling period in the run-up to Christmas and the following Boxing Day and January sales, Amazon needed to translate into English the descriptions and features of approximately 16,000 products already available in Germany, making the products suitable for inclusion on the UK site. The high volume – a word count of over 2.5 million – was the reason why Amazon sought a machine translation approach for this project.
The first challenge was the turnaround time: they needed the majority of the products completed and posted on the site within a three-week window. Secondly, the content was pulled from a variety of data sources, which presented a disparate mix of encoding and formatting challenges. Corrupted characters, incomplete bullet points and HTML tagging inconsistencies needed to be fixed in order to generate text that could be processed correctly. Finally, some of the best-selling product descriptions had to be re-written entirely so as to be relevant to the UK market.
What originally looked like a “straightforward translation task” quickly required technical expertise as well as copy-writing experience, all of which had to adhere to Amazon’s strict guidelines, while respecting their compliance team’s list of banned terms.
During the key selling period in the run-up to Christmas and the following Boxing Day and January sales, Amazon needed to translate into English the descriptions and features of approximately...
With our linguistic and technical expertise we were able to make the content uniform and ready for translation. We developed custom utilities to cater for the Amazon-specific encoding and formatting issues, and to carry out essential HTML code tidy-up, such as automatically adding closing tags and converting HTML entities where required by the Amazon UK system.
The content was converted into a format that presented the key contextual information for every product in an intuitive format, removing the time-consuming need to constantly consult reference material.
For dealing with large quantities of product data quickly, we used our advanced machine translation system, MT plus, a world-leading hybrid solution combining the benefits of statistical and rule-based machine translation approaches. We were able to incorporate our powerful translation memory technologies for rapidly leveraging repetitions and fuzzy match content from Amazon’s own translations, thus achieving the highest possible cost, quality, and time benefits possible for our client.
Finally, for copywriting and post-editing, we put together an expert team of linguists with specialist retail and e-commerce knowledge, and who were specifically trained for post-editing machine translated content.
Quick turnaround, due to the advanced capabilities of MT plus – meaning that what originally seemed like an impossible volume of content was ready for their key selling period
Increased cost efficiencies and high-quality content, as a result of our ability to incorporate translation memories into our machine translation process
Flexible processes to manage the volumes of data and the required deadlines, by including additional proofreading where necessary based on the type of content
From the blog
Our clients say
"Thanks to you and your team for the excellent work and quick turnaround."
John Harrison, Road Marketing Manager
The Met Office