Side by side
| Data Import Wizard | Data Loader | |
|---|---|---|
| Where it runs | In the browser, inside Setup | Desktop app (Windows, macOS); command line on Windows only |
| Records per job | Up to 50,000 | Up to 150 million with Bulk API 2.0 |
| Objects | Accounts, contacts, leads, solutions, person accounts, campaign members, custom objects | Any object, including opportunities, tasks, events, EmailMessage |
| Not supported | Opportunities, cases, campaigns, contracts, products, assets, documents | Nothing at the object level; field-level security still applies |
| Matching | Built-in matching for contacts, leads and accounts (Salesforce ID, name, email, external ID) | Upsert on an external ID field |
| Files | CSV, up to 100 MB | CSV; large jobs split into batches |
| Created dates | Not supported | Supported when “Set Audit Fields upon Record Creation” is enabled |
What the native tools do / where they stop
Both tools load files into one object at a time. Neither one reads from your old CRM, transforms values or keeps parent and child records linked across objects. You do that in the CSVs. Salesforce’s own guidance says to go to a partner or an AgentExchange product for anything larger than Data Loader handles.
What the docs don’t spell out is the load order. A migration is a sequence of Data Loader jobs, and each one depends on the IDs the previous one returned:
- Users (created in Setup, not loaded as data) and any record types.
- Accounts, upserted on a
Legacy_ID__cexternal ID field. - Contacts, with the account looked up through its external ID.
- Opportunities, then opportunity contact roles.
- Tasks and events, with
WhoIdandWhatIdresolved from the success files. - EmailMessage records (insert only via the API) and their EmailMessageRelation rows.
- Files and attachments, last.
If you skip the external IDs, every step means matching record IDs by hand in a spreadsheet. A rerun then creates duplicates.
Rules of thumb
- Under 50,000 contacts or leads, no history: the Wizard is fine.
- Opportunities, activities, custom objects or more than 50,000 rows: Data Loader.
- You need the original created dates: Data Loader with audit fields, and only on insert.
- Several source systems, several million activities, or attachments: script it against the Bulk API, or use a migration tool.
Where FlitStack AI fits
FlitStack AI loads Salesforce through its APIs in the right order: external IDs, lookups and audit fields included, with a sample run and rollback. See the Salesforce Data Loader guide for doing it yourself.