Metadata deployment versus data loading tools in a rollout
Key Takeaway
Deploy what you build in Setup as metadata with change sets or other metadata deployment tools. Load records as data with CSV Data Management. Move the metadata to the target org before the data that depends on it.
An Agentforce Manufacturing rollout moves two different things into the target org. Metadata is the org's setup and customizations. Data is records, such as products, assets, and sales agreements. Each needs its own tool. Change sets and other metadata deployment tools move customizations between orgs, while CSV Data Management imports records from CSV files.
Metadata versus data
To pick the tool, ask whether the change is something you build in Setup or a set of records.
If the change is something you build in Setup, such as a new object...
Deploy it as metadata with a change set or another metadata deployment tool.
If the change is a set of records, such as sales agreements coming from an external system...
Load it as data with CSV Data Management, which you use through the CSV File Import app.
Metadata must reach the target org before the data that depends on it. For example, records for a custom object need that object and its fields in the target org first.
A farm equipment manufacturer is rolling out Agentforce Manufacturing. In a sandbox, the admin creates a custom object called Dealer Region with two custom fields, Region Code and Regional Manager. She built these in Setup, so they are metadata, and she deploys them to production with a change set. Next, she has 8,000 dealer region rows exported from an old planning system as a CSV file. Those rows are records, so a change set can't carry them. After the change set is deployed, she imports the file with CSV File Import and maps Region Code and Regional Manager to the matching column headers.