A Salesforce data backup is an automated copy of your records, files and metadata kept outside the live org, plus a restore you've rehearsed. This guide is for the admin or architect who'll be the one running that restore.
It's 4:50 on a Friday when a sales manager messages you: half her pipeline shows the wrong owner, and a couple of hundred Contacts have vanished. If a Salesforce data backup exists, you lose an hour and some patience; if not, kiss the weekend goodbye.
Each data protection practice below points to the actual setting or API involved. The restore walkthrough near the end puts them to work on a bad Data Loader job.
Why do you need your own Salesforce data backup?
Salesforce backs up its platform for its own disaster recovery. Your org's data is another matter, because under the shared responsibility model that part is yours.
So when someone on your team makes a mistake, the data loss or corruption is yours to fix, and so is the dent in business continuity. Our post on the shared responsibility model draws that line in more detail.
If you're still deciding what a backup has to include, start with our guide to what a backup solution should cover. We'll keep this one narrow, sticking to backup best practices you'd only bother with for Salesforce data and metadata, with its odd exports, API limits and quirks.
Which backup practices are specific to Salesforce?
None of this needs a project plan, since most of it is one setting in Setup or one API call. Skim the table, then dip into whichever rows make you wince.
| What to do | Where it lives | What goes wrong without it |
|---|---|---|
| 1. Schedule exports with files | Setup > Data Export, with the images, documents and attachments box ticked | Lost records and files have no copy outside the org |
| 2. Back up metadata | A Metadata API retrieve driven by a package.xml manifest | A deployment kills a field or a flow and you can't roll it back |
| 3. Plan around API limits | Bulk API 2.0 and the daily API request pool your integrations share | The backup dies halfway, or integrations run out of calls by lunch |
| 4. Know the Recycle Bin window | The Recycle Bin, and Data Loader's Export All | Records leave the bin before anyone notices the delete |
| 5. Keep a change history | Setup Audit Trail and Field History Tracking | Nobody can say what changed, or when |
| 6. Restrict export and restore access | Who holds Modify All Data, and where export files end up | Your backup becomes the easiest way to leak the org |
1. Schedule exports with files included
In Salesforce Setup, find Data Export and choose Schedule Export. Salesforce Help says scheduled exports run weekly for organizations on Enterprise, Performance and Unlimited, and monthly on other editions except Database.com.
Tick both "Include images, documents, and attachments" and "Include Salesforce Files and Salesforce CRM Content document versions," or the contracts and screenshots your team attached won't be there. Each ZIP tops out near 512 MB, and our export guide has the remaining quirks.
Those backup files vanish 48 hours after the email, so automate the download. Park the Salesforce data backup outside the org, the whole idea behind the 3-2-1 backup rule.
2. Back up metadata through the Metadata API
Records are only half the org. The other half is metadata, meaning custom fields, flows, validation rules, page layouts and permission sets, all of which the Metadata API deploys and retrieves.
A typical metadata backup pairs a package.xml with a scheduled Salesforce CLI retrieve. Committing each one to version control lets you diff last Tuesday against today, as our metadata backup guide shows.
Take a data and metadata backup before any big deployment or sandbox refresh, too. If a release deletes a field, the field must exist again before its data can return.
3. Plan backups around your API limits
Third-party backup solutions read your Salesforce data through REST, SOAP or Bulk API. Each call comes out of the same daily API request pool your integrations use.
The pool starts at 100,000 calls every 24 hours, plus an allowance per license. Salesforce says jobs over 2,000 records are good candidates for Bulk API 2.0, which covers most large objects.
Schedule regular backups overnight and watch API usage for the first few weeks. A backup that pushes your ERP sync over its limit at 9 a.m. is hard to explain.
4. Know what the Recycle Bin and Export All can recover
Deleted records sit in the Recycle Bin for 15 days by default, or 30 if Salesforce Support activates Extended Recycle Bin Retention. A restore brings back the same records by ID, plus the child records deleted with them.
A child record deleted on its own has to be restored on its own, and Salesforce only restores lookups nobody has replaced. Look through related lists afterward for gaps.
To list soft-deleted records in bulk, use Data Loader's Export All, not plain Export. It includes soft-deleted records, so run it the moment anyone reports a mass delete.
5. Export change history before Salesforce drops it
After an incident, someone usually asks "what changed?" first. Setup Audit Trail answers that for configuration, but it only keeps 180 days.
Field History Tracking covers up to 20 fields per object for 18 months in the org (24 via the API). Field Audit Trail (sold with Shield or as its own license) keeps it until you delete it, so export both and give the 20 slots to fields you'd hate to rebuild.
Backups holding personal data fall under the GDPR's storage limitation principle too. They need their own data retention rules, and our retention policies post walks through setting them.
6. Restrict who can export and restore
Open an export ZIP and you'll find plaintext CSV files, even for fields Shield Platform Encryption protects in the org. That's sensitive customer data, in the clear, wherever the file lands.
Give backup files the same data security as production: secure storage, encryption in transit and at rest, and access for backup owners only. Restores need tight access controls too.
Any Salesforce user with Modify All Data can restore records deleted by anyone. Keep that group small, check it in your quarterly security reviews, and log every export and restore.
What does a restore look like after a bad data load?
Say it's Tuesday afternoon, and an admin runs a Data Loader update to tidy up Account owners. The column mapping has an off-by-one error.
Territory names land in a text field on 8,000 accounts. A cleanup step in the same batch also deletes 600 Contacts that looked like duplicates, and nobody notices until Thursday.
- Stop the spread. Pause the job and any flows or integrations that copy the bad values into other objects or systems.
- Scope the damage. Use the job's input file, field history on the affected fields and Setup Audit Trail to list which records, fields and settings changed, and when.
- Bring back the deletions first. The Contacts are still in the Recycle Bin window, so restore them there with their original IDs.
- Pick a restore point. You need account values from before Tuesday's job. The last backup, a Sunday export, is only safe for accounts nobody edited on Monday or Tuesday morning.
- Rehearse in a sandbox. Load the backup data into a sandbox and compare the result with production before touching the live Salesforce environment.
- Restore only the damaged field. Run an update keyed on record ID, never an insert, so IDs, data relationships and every other field stay as they are.
- Check and record. Compare record counts and spot-check accounts and their contacts for data integrity, then log what you restored, from which backup and who ran it.
With only a weekly data backup, step 4 means checking by hand every account edited since Sunday. That gap is your recovery point objective, so set it before picking a backup frequency.
Which restore options keep records and relationships intact?
How you restore data decides how much cleanup follows, and whether data integrity survives it. The table shows how each native and third-party backup and restore option copes with Tuesday's mess.
| Restore option | Deleted records | Overwritten fields | Watch out for |
|---|---|---|---|
| Recycle Bin | Within 15 days by default, or 30 with Support | Not covered | Children deleted separately don't return with the parent |
| Export files | Re-inserted as new records | Updated from the last export, keyed on record ID | Re-inserted records get new IDs, so lookups need re-linking |
| Salesforce Backup & Recover (formerly Own) | Restored from daily backups | Restores only the objects, fields and records that changed | A paid add-on for Enterprise, Unlimited and Developer editions |
| Flosum Backup & Archive | Point-in-time restore, per Flosum | Field, record or full-org level, per Flosum | Hosted by Flosum or in your own cloud |
The CSV new-ID problem is covered in our recovering deleted records post. Backup & Recover backs up data, files, metadata, managed packages and sandboxes daily, and its product page adds on-demand backups.
When comparing third-party tools, have each vendor run three restores in a live org. One field across many records, a deleted record with its children, an object back to a set time.
Our buyer's guide covers the other key features to check. These range from metadata coverage and per-object backup frequency to API use and where the backup data lives.
How often should you test a Salesforce backup restore?
Rehearse this scenario in a sandbox before real data loss forces you to. Then borrow the schedule in our HIPAA backup checklist: monthly single-record restores up to a yearly full-org drill.
Time each drill and write the steps into a runbook. Our guide to testing a disaster recovery plan covers the drill, and your targets come from your Salesforce backup and recovery strategy.
Is your Salesforce backup plan following best practices?
Answer each question with a plain yes or no. Any no points you back to the section that fixes it.
- Do scheduled exports include files, and does a script download them before they expire?
- Do you retrieve metadata on a schedule and commit it to version control?
- Do backups run outside business hours, with API usage watched?
- Can someone run Export All within the Recycle Bin window after a mass delete?
- Do you export Setup Audit Trail and field history before they age out?
- Are backup files encrypted, with export and restore access limited and logged?
- Do you back up before every big deployment and sandbox refresh?
- Have you timed a sandbox restore drill against your recovery point objective?
Where does Flosum Backup & Archive fit?
Flosum Backup & Archive runs in Flosum's cloud or in your own AWS, Google Cloud, Azure or private cloud. Flosum says it does granular, point-in-time recovery at field, record or full-org level.
In Tuesday's scenario, that replaces the hand-built update file in step 6. Flosum also says it runs scheduled, incremental and high-frequency backups, and archives older records to storage you control.
Moving inactive records out of the live org keeps reports and searches fast, as our archiving guide explains. You can still pull archived records back for audits and legal holds.
If you're still shortlisting Salesforce backup solutions, put your top two side by side in a full sandbox, with your own data, before anyone signs.
FAQs
What is the best way to protect Salesforce data?
Keep a Salesforce data backup of records, files and metadata outside the org, often enough to match the data loss you can afford. Then prove it with a one-field restore.
Does Salesforce back up my data for me?
Salesforce backs up its platform for disaster recovery, but under the shared responsibility model data is your job. It gives you scheduled exports, the Recycle Bin and paid Backup & Recover.
How often should you back up Salesforce data?
Often enough to meet your recovery point objective, the data loss you can live with. Exports run weekly at best. Backup & Recover runs daily backups and offers Continuous Data Protection for fast-changing data, and third-party tools vary.
Does a Salesforce export include metadata?
No. It exports record data and optional files, so metadata needs its own Metadata API backup, as our metadata backup guide explains.
How long does the Recycle Bin keep deleted records?
15 days by default, or 30 if Salesforce Support activates Extended Recycle Bin Retention. It won't help with fields that were overwritten instead of deleted.
Why does restoring from a CSV break relationships?
Re-inserting from CSV files gives each record a new record ID, so lookups pointing at the old ID must be re-linked. For overwritten fields, update on the existing ID instead.
Do Salesforce backups use API calls?
Yes. Any backup reading through REST, SOAP or Bulk API uses the same daily API request pool as your integrations, so run big jobs after hours.
Book a demo to see a field-level restore after a bad data load in Flosum Backup & Archive.
Thank you for subscribing




