For an international eCommerce application, the first logical capacity is the ability to do things like translate all of the labels, static texts, maybe the menus and even some of the product information.
Perhaps it won’t be deemed necessary to translate literally every speck of content on the site. You might rather leverage an API to translate the majority of the content, without the user having to use Google Translate, so that they would get more of a seamless experience.
Actual human translation may only be used for the standard calls to action, the buttons and some of the key application components. Those require a high degree of adherence to each language, dialect and local context within the international eCommerce platform.
The main idea is that you go after the most critical pieces of the translations, where users are going to get the most traction. But you deliberately choose not to do all of the “more detailed”, possibly low payoff, translation work for everything else.
Instead, you opt to phase from a Google Translate type of implementation, where the user is interacting with a relatively clunky but technically functional translation, to a far superior standard of translation for key items. The following are examples of content types that could be translated professionally:
- Buttons
- Calls to action (CTA)
- Menus
- Titles
- Important pieces of product and category data
The remaining content, such as the text within product descriptions, could be lengthy and difficult to constantly translate as new products come in. This kind of content can just pass through an API for either Google Translate or an alternative mechanism, which repeats the process every time a product gets modified.
After passing through the API, the content gets automatically updated for each language. There’s no human involved, so nobody is specifically looking at all of the nuances for that language, dialect or region. But the outcome is still functional in respect to the detailed content and also very high fidelity within the buttons, calls to action, labels, tool tips, etc.
As mentioned before, a lot of the product data itself could be manually translated as well, like the product titles. Such practice has potential to get very far because a lot of people are simply going to look at the imagery.
A large percentage of users view some of the title information, and then they quickly skim through the content; they're not necessarily reading it in detail. From a multilingual capability aspect, manually translating key elements that don't cost nearly as much as doing everything can make the platform much more competitive.