How AI Can Shorten Time-To-Market in Multilingual eCommerce

Multilingual eCommerce site

Creating product content in multiple languages

According to Internet World Stats, only 26% of internet users speak English.

That is why, in order to reach broader audiences online with your products, you will usually require setting up a multilingual eCommerce site that serves customers from different countries or language areas.

However, translating your website into different languages can entail some challenges. You will not only need to translate all sections like the menu or the home page, but you’ll also need to localize the payment processes, invoices or purchase orders, etc. In short, anything that you display in your site.

Moreover, if we think about how important product information is in eCommerce, we soon realize maintaining an eCommerce site with every product sheet translated into several languages is time-consuming and quite complex.

Every single touchpoint in your customer’s journey will need to be adapted to fit different markets. Of course, once the work is finished, if well done, it will translate into a huge increase in traffic and sales.

Average eCommerce translation spending

ecommerce translation spending

Delivering all content in different languages requires translation, which is quite costly. Although your eCommerce will have some static content, you will have to address every change within your catalog. The most dynamic content in that catalog will be your product pages.

A human translator (or what’s more common a translation company) would typically ask retailers for 0.10€ to 0.12€ per translated word. Product specifications, usually in the form of attributes, contain, on average, up to 12 words. Then, in some cases, you may have some written descriptions that add up to that number, reaching the 100 or 150 words.

This means that if you want to make your eCommerce site available in 5 languages or countries, and you have a catalog made of 4,000 products, you can easily spend between 21,120€ (for product specs) and 264,000€ (for whole written descriptions) in translation. Even with large-volume discounts of 3% to 5%, these are still big numbers. This is a task you would have to repeat every season, every time the product range is updated.

Attribute translationTranslation of descriptions
4,000 prod x 12 w = 48,000 w4,000 prod x 150 w = 600,000 w
48,000 w  x 0,11€ = 5,280 €600,000 w x 0,11€ = 66,000 €
5,280€ x 4 lang = 21,120 €66,000€ x 4 lang = 264,000 €

*We assume 0,11€ as the translation rate in both cases.

Impact on operational processes

operational processes

Seeing the magnitude of such a project, it becomes evident that making your shop available in different languages will take a whole lot of time and imply a huge eCommerce translation spending!

Usually, a human translator would be able to translate between 2,000 and 3,000 words a day. In the case of an eCommerce site that only provides product attributes, we would have a translation project of around 48,000 words per language.

If we assume that 4 different translators can work at the same rate, translating the content into the 4 different languages, this would account for at least 16 days of translation. If you work with a translation agency, the period could be easily extended with proofreadings, revisions…

Additionally, what is also critical is terminology. It is imperative that you make sure to use the terms matching your eCommerce tone of voice season after season. Keeping consistency can be difficult when outsourcing translations.

The cost of product data mistakes

Due to their complexity, the above-mentioned tasks are prone to mistakes. There are high chances of displaying wrong or mistaken information. In fact, 30% of data in eCommerce sites have at least one error. Spotting mistakes after translation is a difficult task, and so is correcting them. The cost of amending mistaken data is estimated between 60 € and 80€ per product.

This is why unifying product information and being able to manage all languages from a single platform is essential. Holding control over the content will help you provide consistent data and reduce the risk of mistakes.

Reducing eCommerce translation spending through AI-generated content

AI-generated product content

Retailers have started to understand the importance of avoiding the use of quick fix automatic translation apps like Google Translate.

Google and other translation apps use already translated and published content, as the source to provide a version of the text you input in another language. This, of course, prevents it from delivering an adapted translation that fits your specific style and tone of voice.

An example of how to solve this problem of translating product attributes is with Artificial Intelligence solutions like the one Catchoom offers. Solutions like this are able to automatically extract product attributes in several languages in an automated way by analysing product images through Computer Vision.

Catchoom eliminates the need for manually translating each attribute for every single product when a new collection or piece of clothing is uploaded to the eCommerce platform.

Instead, the category tree and all specific product attributes in the original eCommerce language are mapped to the corresponding attributes and categories in the rest of the languages. This means you only need to translate product attributes and category names once. Subsequently, Catchoom will analyze every product image uploaded to the system and will automatically output the data in the different eCommerce languages.

This saves a lot of time and speeds up time-to-market by reducing the need for translating product content repeatedly. And what’s more, it saves capital.  If you want to dig deeper and see how it translates into a good ROI, don’t miss these infographics.