E-Commerce and Retail

The second determining factor is that the content is always changing. Prices are updated, products are added, campaigns start and finish. A one-off translation project begins to date the day it goes live.

The catalogue: where volume is managed

Product text is heavily repetitive. The same material, the same care instructions and the same size chart run through hundreds of products. Managing that repetition with translation memory is the single largest factor in the total cost of a catalogue project. Where there is no memory, the same sentence is translated a hundred times and paid for a hundred times.

The second technical issue is the structure of the product data. If attribute fields are made up of defined values rather than free text, the translation is done once and used across the whole catalogue. Fields left to free text mean separate work for every product. A localisation decision therefore usually begins with a review of the catalogue structure.

Terminology risk and brand voice

In retail, terminology risk shows up in two different places. The first is in product attributes: getting the name of a fabric, a technical feature or a size measurement wrong pushes the return rate straight up. When sizes and measurements are carried across without being converted, the customer receives the wrong item and the cost lands on logistics.

The second is brand voice. Forms of address, the level of familiarity and the use of humour vary from market to market. A playful headline that reads naturally in Turkish may look flippant in another market. Drawing up a short style guide before a catalogue is translated costs less than the bulk corrections that would otherwise follow.

The third and most expensive risk lies in the legal text. Return periods, the right of withdrawal and warranty terms are subject to the consumer regulations of the target market; translating them word for word can create a commitment that was never intended. These texts are less a matter of translation than of review against the target market.

Compliance and the data side

In e-commerce text the need for confidentiality is not as sharp as in law or healthcare; the catalogue is public in any case. Pre-launch product information, price lists and the campaign calendar, by contrast, are commercially confidential and have to be protected until the publication date. Writing the embargo date into the contract is a practical necessity.

With customer reviews and support correspondence, the personal data dimension comes into play. Review text can contain names, addresses and order numbers; stripping these fields out before the text goes for translation is the right approach both for compliance and for cost.

Marketplaces work by rules of their own

For brands that sell through international marketplaces rather than through their own site, the rules change. These platforms impose strict formats and character limits on product titles, bulleted feature lists and description fields. The translation has to stay inside those limits; a title one character over will not go live.

The second difference lies in how search works. On a marketplace the product is found through the platform’s own search engine, and that engine usually gives heavy weight to the words in the title. Translating a title therefore stops being a linguistic exercise and becomes a matter of placing the right keywords within a limited number of characters.

The third difference is category mapping. The platform’s own category tree may not line up with the brand’s own classification. A product placed in the wrong category stays invisible to the intended customer, however good its translation.

Customer service and returns correspondence

The most commonly overlooked consequence of starting to sell in a new language is that the questions come in that language too. If the catalogue has been translated but the support side has not been prepared, unanswered messages pile up in the first weeks and review scores fall. This is the typical bottleneck many brands hit on their first move abroad.

The practical answer is to translate the questions that come up most often in advance, as a library of ready replies. Headings such as delivery times, return conditions, size exchanges and invoice requests are much the same across all markets. Once that library is in place, the support team can give consistent answers without speaking the language; only the exceptional cases go for translation.

In returns correspondence the tone of the language becomes critical. If consumer expectations in the target market are closed to a bluntly negative style, a Turkish template translated word for word leaves a poor impression. These texts should be rewritten for the target market rather than translated.

Measurement and feedback

In e-commerce, translation quality can be measured directly, and that is the sector’s greatest advantage over the others. Conversion rates by product page, the reasons given for returns, on-site search logs and the steps at which baskets are abandoned all show where the text is stumbling.

The most valuable data is the reasons for returns. If returns cluster around “it was not the product I was expecting”, the problem usually lies not in logistics but in the product description. In the same way, words that are typed often into on-site search and return no results show that your category names have not settled into the language of the target market. Read regularly, these two records put translation revisions on a footing of measurement rather than guesswork.

The third source of measurement is the questions reaching the support team. If the same product feature draws question after question, that feature’s description is not being understood in the target language. This feedback makes it possible to correct only the problem areas instead of reviewing the whole catalogue.

How we work

The model that works in e-commerce is a continuous arrangement, close to a subscription. The first transfer of the catalogue is run as a large project; after that, a regular rhythm is set up for new products and campaigns. Weekly or fortnightly delivery cycles make it possible to work in step with the campaign calendar.

What decides the cost is the groundwork laid during that first project: the termbase, the style guide and the translation memory. Once those three are in place, the unit cost of new content falls a little further each period. Depending on which markets you are opening up in, you can work out a language plan from the language pages, look at the volume tiers on the pricing page, and share your catalogue inventory through the quote form.