When switching to a new ecommerce platform or to another system, such as a marketing automation tool, merchants often overlook one key component: the data. Most providers will claim that the data migration is easy. But, in my experience, that is not always the case.
I’ve managed many data migrations, small and large. In this article, I’ll provide a data-migration checklist that is safe, efficient, and accurate. Certainly loss of data is the primary risk. However, there are other data concerns to pay attention to.
Back up the data. Before starting the migration, back up your data. Make sure that the backup can be accessed at any point in the future. Date stamp your system periodically so that you can easily reference your data should anything go wrong. This is good practice even if you are not migrating.
Assemble a team. Before migrating, assign roles to your team. Who handles what data? Which fields do they access? Defining the team roles and data usage of each can clarify responsibilities and decisions.
Develop a data migration plan. Once your data is safely copied, document your data migration plan. Start by identifying the data you are migrating and where it resides. Map out all your hierarchies and dependencies. You may decide to bring only partial data.
Audit the data. Regardless of what you are migrating, audit the data beforehand. This is a good time to clean it. Look for the following items.
- Duplications. Remove duplicates.
- Data completeness. Incomplete data can be unusable.
- Outdated records. An outdated record could be a bounced email address or a previous name.
- Junk (test) records. Developers often test the system with hypothetical records. Remove them before migrating.
Identify fields and tables. Identify what fields and tables you are migrating. During a data audit, it is common to learn that, say, 300 fields exist in the database, but only 20 or 30 are in use. A quick test is to check the percentage of a specific field that contains data.
Regardless of what you are migrating, audit the data beforehand.
Another way to narrow down the data is to speak with team members to identify the data’s use cases. For example, customer service personnel may not care about the total lifetime value of a customer, but the sales staff could. Assigning key fields for each team member can help determine the use. Remember that some fields could be connected to multiple data tables. Thus, assemble a data map of all records, fields, and tables that are affected.
(Data terminology can be confusing. Records are comprised of fields, each of which contains one item of information. A set of records constitutes a table. For example, a customer table might contain records that include three fields: a name field, an address field, and a phone number field.)
After you’ve backed up the data, created a migration plan, performed an audit, and identified fields and files, you should know which records to migrate and which to keep only in your backup file. Consider creating a field that designates if a record is migrating, or not. Each business is different. Which records add value to the new platform and which simply clutter it?
You are now ready for data migration.
Prepare a data map. When migrating your data, create a map of your old data to its new location. Pay attention to fields and their formatting. Date fields, for example, may not migrate correctly if the old platform used YYYY-MM-DD and the new system is DD/MM/YYYY.
Check for dependencies. If there are data dependencies, make sure the lookup has been established before loading the data.
Verify. Before going live on the new platform, verify that the migration was successful. Double-check your data.
- Make sure all records flagged for migration have been migrated.
- Confirm that the total number of records on the new platform is correct. Search for a few records to make sure they exist in the new system.
- Ensure that all fields in the new platform are in the proper format. To do so this, download a sample of 20 records with all fields and compare to the original data.
- Finally, make sure all dependencies are still there.
Train users. Now that you have successfully migrated, train all users on proper data usage. Remember the four Cs for healthy data: complete, current, correct, and consistent.