DHL’s Cross Country Challenges for Data Consolidation
DHL is the world's leading express and logistics company offering customers innovative and customised solutions from a single source. With global expertise in solutions, express, air and ocean freight and overland transport, DHL combines worldwide coverage with an in depth understanding of local markets.
DHL continues to be at the forefront of technology and with more than 150,000 dedicated employees and an international network linking more than 220 countries and territories, it
guarantees fast and reliable services, aimed at exceeding customers’ expectations.
When DHL International (Singapore) decided to move to a new Customer Relationship Management system integrating countries in the Asia Pacific region, the first issues that need to be addressed were those of data migration and data de-duplication. “We wanted to start with a de-duplicated set of client records for each country” said Henry Chin, Business Systems Manager, Asia Pacific. “Our existing systems do not give us easy visibility across multiple customer records created over time, and there were many differences in the records due to typos, abbreviations etc”.
In December 2002, after some initial research, the Commercial Research and Planning Division approached Search Software America (SSA) and were immediately impressed with the number of countries and languages that its systems supported. With the exception of South Korea Romanised (which is not supported in the original SSA population rules), all of the other countries involved in this project were supported. (South Korean language support was subsequently delivered by SSA within a month.)
Given the tight time frame for the project, SSA set up a fully-functional demonstration of the Data Clustering Engine (DCE) and illustrated its ability to match and group records using data provided by DHL. Convinced of the software’s ability to match the data with great accuracy, the group purchased DCE v2.2 and invested in on-site training.
By January 2003, the Data Clustering Engine was running in production on a Windows 2000 server and processing data in Singapore from a sub-set of the required countries, including Singapore, South Korea, Australia, Taiwan, China, Indonesia, Thailand, India, and Japan.
While the project is still in progress, (not every country’s data has yet been processed and the task of consolidating the duplicates, once identified by DCE, is still ongoing), the specialist support from SSA, the training, and the ease-of-use of the software made it possible to get this part of the project up and running in a very short period of time.
“The Data Clustering Engine has allowed us to better identify duplicate records, leading to better allocation of sales resources”, said Chin. “We are extremely pleased that up to 95% of duplicates have been identified in the runs so far. This project will only run from within Singapore but the benefits of improved data quality will be felt across the region in more than thirteen countries.”
Chin added, “Search Software America has provided a variety of support to our team. They continue to provide advice in getting the optimum results from the DCE product and have given us updated rule-sets promptly as we identify data that requires specific treatment. The population rules in DCE can be fine-tuned easily, allowing our users to explore different scenarios and see the results quickly.”
According to the Research & Planning Manager, “DCE is easy to use, the controls are intuitive and a new user can be brought up to speed within a couple of days.” The division’s Marketing Information Manager added “DCE provides reliable clustering results, reducing the amount of manual validation required by at least 70%.”
(Based in Brussels, Belgium, DHL is 100% owned by Deutsche Post World Net. For more information visit www.dhl.com)
|
The Commercial Research and Planning Division at DHL Asia Pacific focus on the data was primarily on organisations. All data captured in the various countries is in romanized form subject to local conventions. Actual data volumes vary per country, up to around 650,000 records in some cases.