In most computer systems the term transliteration is used in the context of converting from a non-Latin alphabet to the Latin alphabet, or Romanization. In the real world, however, transliteration can occur between any two alphabets.

For example, a United States organization with offices in the US and Japan decides that all of its Japanese customer data should be captured in Japanese and in Romanized form to maintain a single language view of the corporate databases. A bank in Saudi Arabia captures customer data in Arabic for local needs, and in English to satisfy needs for inter-bank wire transfers and compliance regulations.

Transliteration may be done formally (conforming to a documented standard – although there will often be a number of standards used by different groups or organizations); or informally (by ordinary people in their normal day to day work, adding personal interpretations to the mapping choices and frequently changing the rules and making mistakes.)

Different formal transliteration standards and informal transliteration may co-exist in the same system/ database, and result in significant variation in the transliterated form. Transliteration also has an attribute of direction. Forward transliteration refers to transliteration from a name’s original script to a target script (e.g. "Romanization" of an Arabic name from Arabic to English; "Arabicization" of an English name from English to Arabic.) Reverse transliteration refers to the transliteration of a name from its representation in a foreign script to its original script (e.g. "Romanization" of an Arabic name recorded in English back to Arabic; "Arabicization" of an English name recorded in Arabic back to English.)

In addition to data recorded in a local script, a system/database may contain data that has been the subject of any combinations of formal and informal, forward and reverse transliteration.