The purpose of this blog series is to describe the top 12 steps to ensure successful software implementation. The first nine steps, Planning, Process Design, Solution Design, Configuration, Customization, Integration, Reporting, Training and Testing were discussed in my previous blog posts.
This post is about Data Conversion. A successfully run data conversion can make your project. An inaccurate or incomplete data conversion can destroy it.
If I had to provide only one best practice piece of advice on data conversion, I would say: “Only convert what you need.” If you minimize your data conversion and are still able to run your business on a new system, then leave it behind. Data conversions are time-consuming and costly and, even with extensive effort, they can still fail.
To get started, review your data and determine what you really need. Normally, the conversion of both master records and transaction history is considered.
Consider converting only the master records that were recently active. For example, converting 1000s of customers with whom you have not interacted in years is seldom valuable and may require you to clean data records that you may never use in your new system. Consider converting only master records that have been used in the last 12 – 24 months. It is normally easier to map historical master records to a new system then transactional data, but it is essential that the business owners of the data participate in the mapping. If they do not, it should be reasonably expected that mappings will be incorrect and multiple iterations will be required to successfully complete the conversion.
Historical transaction data is often very difficult to map to new systems. Normally, new systems are put in place in order to transact differently than in the old systems. This means that there may not be any actual correlation between your old data and your new data. Adding complexity to capture your old data in your new system may be counter-intuitive.
Some tips for a successful data conversion are:
- Start early. Include the person responsible for the conversion in the design so they can ensure that necessary data fields exist.
- Document mapping of the old data to the new system so it can be reviewed and approved by the business users.
- Try each conversion point first by just manually entering the data. Take a record from the old system and enter it to the new system. This will ensure that the mapping actually makes sense.
- Start with a small sample set of data from each larger set of data. Before you need to convert hundreds or thousands or even millions of records, be sure you can convert 20!
- Look at the data in the user interface after it is converted. It needs to be correct and make sense from the user’s point of view.
These points are just the tip of the iceberg. Be sure to have team members with expertise on your team to do the data conversion. Experienced data conversion specialists will ensure that you avoid common oversights or errors.
Stay tuned for the final two blog posts in this series: Deployment and Project Management.