Salesforce gives you two free import tools, and MuleSoft sells a third. They overlap, but each has a clear job. The short version: the Data Import Wizard is for small, simple files, Data Loader is for volume and for any object, and dataloader.io is Data Loader’s browser-based cousin with monthly caps. None of them is a migration tool. They load rows you have already exported, cleaned and keyed. Everything else is your work.
Which Salesforce import tool should I use?
Salesforce’s own guidance splits the tools by volume, object and how often you load. Here is the decision in one table.
| If you need to… | Use | Why |
|---|---|---|
| Load fewer than 50,000 records into accounts, contacts, leads, solutions, person accounts, campaign members or a custom object | Data Import Wizard | Runs in Setup, no install, built-in duplicate matching |
| Load opportunities, cases, activities or any object the wizard doesn’t list | Data Loader | Supports all objects, including custom objects |
| Load 50,000 to 150 million records | Data Loader | Supports CSV files up to 150 million records with Bulk API 2.0 |
| Run a nightly or scheduled load | Data Loader (Windows CLI) or dataloader.io | Command-line config files, or scheduled tasks in the browser |
| Work from a browser on a locked-down laptop | dataloader.io | Nothing to install. Monthly record caps apply |
| Move a whole CRM, with history and relationships | A migration tool or service | The loaders don’t extract, map or sequence across objects |
Above 150 million records, Salesforce’s docs recommend working with a Salesforce partner or an AppExchange product.
What does the Data Import Wizard do, and where does it stop?
The wizard lives in Setup (search “Data Import Wizard” in Quick Find). Trailhead describes it as able to import up to 50,000 records at a time into standard objects (accounts, contacts, leads, solutions, person accounts, campaign members) and custom objects. You can add new records, update existing ones, or both, and it can prevent duplicates by matching on account name and site, contact email or lead email.
Where it stops:
- No opportunities, cases or activities. If the object isn’t on its list, you need Data Loader.
- Validation rules run. Records that fail are skipped, so a strict org can quietly drop rows.
- Simple mappings only. Salesforce’s guidance is to use it when the target object has fewer than 50 fields and the data has no complex mappings.
What does Data Loader do, and where does it stop?
Data Loader is a client application you install. Salesforce describes it as a tool to insert, update, delete or export records, reading from CSV or a database connection. Key facts from the current documentation:
- Volume: CSV files up to 150 million records when used with Bulk API 2.0. Older versions of the docs said 5 million, so check the version you are reading.
- Objects: all objects, including custom objects.
- Platforms: the interface runs on macOS and Windows and needs Java 17 or later. Salesforce no longer bundles Java with the Windows installer.
- Automation: the command-line interface, which you use for scheduled jobs, is supported on Windows only.
- Editions and permissions: available in Enterprise, Performance, Unlimited and Developer editions. Users need API Enabled, the object permission for the operation, and Bulk API Hard Delete if they hard-delete through the Bulk API.
- Logs: success and error files in CSV for every run.
Where it stops is the part most teams find out mid-project:
- It loads one object per job. To link contacts to accounts, you first load accounts, export the new Salesforce IDs, VLOOKUP them into the contact file, then load contacts. Repeat for opportunities, then for every activity type.
- Created dates and owners need setup. To keep original Created Date and Created By, an admin must enable the Create Audit Fields feature, which lets API tools such as Data Loader set those system fields on insert.
- No undo. Reversing a bad load means building and running a delete job from the success file.
- It doesn’t export attachments. Salesforce’s export docs say to use the weekly export for those.
What is dataloader.io and how is it priced?
dataloader.io is a browser-based loader that shows a Salesforce copyright and is “Powered by” MuleSoft’s Anypoint Platform. It does import, export and delete, connects to Box, Dropbox and FTP, and schedules tasks. Published plans as of November 2026:
| Plan | Price | Records per month | Max file size | Scheduled tasks |
|---|---|---|---|---|
| Free | $0 per user per month | 10,000 | 10 MB | 1 |
| Professional | $99 per user per month | 100,000 | 50 MB | 50 |
| Enterprise | $299 per user per month | Unlimited | 100 MB | Unlimited |
It counts production and sandbox users separately. The same limits apply as with Data Loader: one object per task, and you supply the IDs that link records.
How do I import data into Salesforce with Data Loader?
This is the procedure Salesforce’s Trailhead module describes, with the steps that bite in practice added.
- Export from the old system. Get one CSV per object: companies, people, deals, activities, notes.
- Clean the files. Remove duplicates and unneeded rows, fix errors, and standardise picklist values to match Salesforce exactly.
- Compare fields and map. List each source field against a Salesforce field. Create missing custom fields and picklist values first.
- Prepare the org. Enable Create Audit Fields if you need original dates. Temporarily deactivate validation rules, duplicate rules and automations that would fire on insert.
- Load parents first. Insert users (or map owners to existing user IDs), then accounts.
- Carry IDs forward. From the account success file, map old IDs to new Salesforce IDs, add them to the contact file, then insert contacts. Repeat for opportunities, contact roles, cases and activities.
- Test with a small file. Load a few hundred rows of each object, check them in the UI, then run the full set.
- Reconcile. Compare row counts and spot-check records against the source. Reactivate the rules and automations you turned off.
Steps 5 and 6 are where most of the hours go. A CRM with five related objects and four activity types is nine dependent loads, each needing the previous load’s IDs.
When is a done-for-you migration better than a loader?
The loaders are the right tool when you already have clean, Salesforce-shaped files. They are the wrong tool when the hard part is everything before the file: extracting from another CRM’s API, deciding how its objects map to Salesforce’s, keeping activity history and owners, and proving nothing went missing.
Choose a service when:
- the source is another CRM (HubSpot, Zoho, Pipedrive, Dynamics), not a spreadsheet;
- you need emails, calls, tasks and notes on the right records with original dates;
- there are custom objects or record types to map;
- you can’t freeze the old system for days during cutover;
- someone will ask you to prove every record arrived.
When to bring in FlitStack AI
FlitStack AI is a data-migration platform sold as software plus a done-for-you service. It connects to the source and to Salesforce, proposes field-level mappings with AI for you to review, runs a sample, then migrates every object in order with validation on each record and one-click rollback. A delta migration picks up records changed during cutover, and the Salesforce implementation service sets up pipelines and automation in the new org.
Moving from HubSpot? See the HubSpot to Salesforce migration page, or start from the Salesforce migration hub. For a scoped quote, book a free scoping call.