7 Best Practices for ERP Data Conversion or Migration

I will not talk about the process of data conversion or data migration during an ERP project but would like to share the best practices that will help us achieve a better data quality index in your new system. To start-with, we need to keep our focus clear on what are our success factors or how will we measure that the job has been done completely. Per my view, we have four major success factors or key metrics which are required to be measured at the end and these are:
  • Data Accuracy
  • Data Consistency
  • Completeness of Data
  • Validity of Data
One of my earlier posts provides details on how to measure these metrics. Coming back to the best practices, let me list them down below:
  1. Complete your pre-work: The ingredients of the pre-work are:
    • Identification of source legacy systems & source business units
    • Find out which system is best to provide data on what as there might be duplicate data in systems
    • Identification of data owners, data cleaners and data verifiers
    • Chart out a dedicated project plan for data conversion (data migration)within the ERP project plan with a dedicated owner for this activity
    • Decide on the size and period of data that needs to be converted with business logic (Tip: The thumb rule is to have last 2 years of transactional data and the open transactions in addition of all master data)
  2. Workout Sessions: Plan extensive work out sessions between legacy system owners, consulting vendor team and the data owners to nail down different opinions and approaches and come out with a common approach, format of capturing data and segregation of manual as well as automated steps in data conversion
  3. Clean and Re-clean: No data can be clean enough to be converted to the new system and it is highly recommended to clean the data in terms of not only removing special characters, junk characters etc. but also similar looking names of customers, suppliers, employees, buyers etc.
  4. Plan about the deltas: After you extract the data from legacy systems and do your activities like cleaning, segregating and finally conversion, there will be some time when the transactions are still being done in legacy systems and data is still being created in there which is not part of your data dump. Plan in advance to handle this situation and how would you transfer that small deltas from different systems to the new system
  5. Sequence out the data loads: Identify the dependencies of one data on the other and plan the sequence of conversion accordingly. For example, employees need to converted first and then buyers can be converted. This will not only save time and effort but will also make some automatic validation in the system to improve on data accuracy
  6. Test the cycles twice: Before actual conversion or migration of data, it is strongly recommended to test the conversion run at least twice and validate the data correctness, completeness, consistency and validity as these simulations will bring out potential issues which can ironed out for the final data conversion run to production system. Also, include the load testing of data conversion as part of test runs
  7. Verify the Data: Plan to have random verifiers from the business user fraternity to verify their data in addition of the verification done by the identified data verifiers. This will ensure successful conversion of data
The most important part of the data conversion or migration is regular communication and co-ordination between separate teams (specially if we have multiple legacy source systems) and I have seen in my experience that following the simple rule of sharing ideas, works miracles in data conversion.

No comments:

Post a Comment

Your thoughts are welcome