Localization is the process of modifying content (or a product) to make it usable for a new locale. Often, this includes translating the content from the source language into the language used in the locale. If, for example, you are selling in the U.S. market and want to expand into Germany, you probably need to translate your content into German. The localization process, however, includes more than just translation. For example, U.S. and German currencies are different, so you might need to change references from dollars to euros. It is possible to localize information without translating it—such as changing examples, currency, geographically specific information, and perhaps color schemes.

There are some standard best practices that help to minimize the cost of localization, including the following:

  • Source text should use clear, concise language
  • Do not use jargon, idiom, metaphors, puns, or other creative language
  • Use simple words
  • Use simple language structure

These best practices are appropriate for technical content, especially procedures and reference information, which need to be highly structured anyway. The readers in the source language also appreciate information that is presented consistently. For example, in providing menu instructions, choose one standard, such as File > Open, and use it consistently. Do not use any other variations, such as: choose File, then Open; select Open from the File menu; from the File menu, select Open; or click File, Open.

Following these best practices will help to reduce the overall cost of localization. Translators can more easily understand the content and deliver it in other languages and, importantly, they can use translation memory tools that recognize patterns and support some automation in translation.

The cost of efficiency

Optimizing content for efficient translation means letting go of interesting word choices, pop culture references, and other creative communication approaches. The result is bland sentences, which is acceptable for most technical content. But if your communication strategy requires you to make an emotional connection with the reader, you may need to use more creative language. Be aware that these approaches make translation more expensive—it takes longer for the translators to develop an equivalent phrasing in each target language. For example, the expression “once in a blue moon” cannot be translated literally because “blue moon” doesn’t have a special meaning outside of English. Instead, the translator must think of a metaphor that means “very rarely” in the target language and that fits the tone of the message.

Visual communication

Translating text is challenging, but graphics, video, and other visual components are even more difficult. You can create graphics that are culturally neutral and appropriate for multiple languages, but there are myriad challenges, including the following:

  • The meaning of a color varies across cultures.
  • Images of people are not appropriate in some cultures. In other cultures, images of women are not acceptable.
  • Visual representations of hands or feet are not appropriate in some cultures.
  • It is difficult to design icons that are universally understood.
  • Visual metaphors or puns may not work in translation. (For example, the idea of an owl is a symbol of wisdom in the United States and parts of Europe. But in other cultures, owls are associated with death and witchcraft.)

Automating translation

For technical content, translators generally charge by the word. At 10–25 cents per word (per language), translation expenses rise quickly. For example, a 100-page document would have approximately 25,000 words, so translation would be at least $3,000 per language. (This quick estimate does not factor in project management, graphics, or anything else that might complicate the process.)

The idea of automated translation is deeply appealing—computers can do nearly everything else, so why not this?

There are two major categories of automated translation:

  • Machine translation. This refers to a process where the source language is fed into the system and the target language is generated as output. The quality of the translation is not very good, but it is usually sufficient for the reader to understand the general meaning of the original text.
  • Computer-assisted translation. Translation memory (TM) systems contain a database of previously translated language pairs. When a new translation project is loaded into the system, the TM is scanned for matches. Thus, if you have already translated a sentence (“The database is read-only”) into multiple languages, and that sentence occurs again in new content, the TM system will recognize that there is a match and translate that sentence for you. When you load in a sentence that is similar to previously translated content (“The file is read-only”), the system may recognize this as a “fuzzy match” and provide you with a proposed translation. Human intervention is then required to validate the fuzzy matches and ensure that the final document makes sense in context.

Computer-assisted translation is the current standard and should be used in any translation effort so that the translation is stored in a reusable manner for future use. Most organizations that use professional translation companies have translation memory in place.

Many organizations today are using “in-house” translation. In some cases, they have a professional translation team, but often, the translation process involves shipping the content to in-country employees (often, sales/marketing or engineering staff) for translation. This approach is acceptable for a one-time translation of a small amount of information, but any company with serious global operations needs to invest in a professional translation workflow.

Optimizing your localization process

There are several opportunities to ensure that localization is as efficient as possible.

Table 1. Opportunities for efficient localization
Stage Best practice
Content development

Create clear, simple, concise text.

Limit text in graphics, and put text into an editable layer.

Use legends rather than embedded text for graphics.


Use predictable structure.

Use styles or structured authoring.

Do not use embedded formatting.


Include a language flag on content so the system can deliver content to readers in the language they want.


Use professional translators.

Use translation memory systems.

Work with localization professionals to identify potential problems.

Look for opportunities to change source content instead of requiring changes in each target language.

Add comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.