Data Quality In China

The problem is that it is not easy to merge data that is built using different systems:

• Customer names entered in Chinese characters need to be transliterated to match consumer records with the names entered in other markets. However, transliteration systems give different results, making automatic matching difficult.
• Chinese names entered into Western systems are not always entered in the same way by data entry personnel.  For example, the Chinese surnames Wang, Huang, and Wong all refer to the same surname.
• Address data is often inconsistent.  As mentioned earlier, there are differences in preferred China address formats depending on sources.  A Microsoft-based system using the Microsoft Developer Network format may not match other systems.
• Western systems often do not have all of the needed fields in the proper order to represent Chinese addresses.  For example, in many languages, addresses are usually written from the smallest location to the largest In China, this is the reverse. Here is a common way in which addresses would be written in the Mainland:
  • 伽 (shěng): province
  • 群 (xiàn): county (you don’t need the county for a large city)
  • 懇 (shì): city / town
  • 혐 (qū): district
  • 댕쌍 (dàjiē): avenue
  • 쨌 (lù): road
  • 짜 (lóu): building
  • 杆 (shì): flat / apartment (or 酪 for a house)

When Mr. Li is traveling abroad, he would want to provide more information than the fields provided in a typical Western data entry form.  As a result, some address information may be combined or left out so that it could fit.  The data quality would be poor and would need to be improved.

The challenge for creating a single customer view for a seamless experience, analysis, or forecasting is that data quality parsing and cleansing must be applied consistently for both the Chinese domestic data and the data captured overseas.  A company with both Chinese market experience and broad international experience is required to make it work.