Originally, implementations followed industry-standard cycles. Specifically, we would look at the existing data, normalize them offline and load them using a specific component built for that purpose. But we not only had to bring the data into our global instance, we also had to standardize them to support central data processing hubs or shared services centers. We could not afford managing by exception. We needed short process cycles and we had to be cost efficient. Operating in more than 30 countries with numerous country-specific features, ranging from different languages, different character sets, different number formats and different tax requirements, the data conversion posed a significant challenge.
One such challenge was the creation of new suppliers. Suppliers were created in small batches; the amount of time spent on creating a supplier properly was disproportionate compared to other activities. The specific knowledge it required in any given country to complete all the necessary fields properly created a significant control burden. Not only was the control inefficient, by the time a data entry error was discovered, the implications had snowballed. In addition, supplier-based reviews meant a significant IT effort, and so did mass updates. They were more frequent than anticipated. Tools like Toad could be used to extract the data but could not be used to reload them. Also, they granted a level of access to the data that made the auditors uneasy. Validation was mostly visual by sample, and the process would unequivocally go through an Excel spreadsheet anyway.
Company integrations would take at least six months, with data migration being a significant bottleneck.
Figure 1 depicts a data conversion run programmatically.