If your sandboxes are still masked by the legacy Data Mask managed package, you have about six weeks left. Salesforce Data Mask end of support has already begun: Salesforce’s retirement notice says support ended on 9 July 2026 and the package’s display becomes read-only on 31 December 2026. It also says existing masking configurations are not converted to Data Mask & Seed, the core app that replaced the package in Summer ’26. Every configuration has to be rebuilt.
For what to mask in the first place, see our guide to Salesforce data masking for sandboxes.
What the two dates mean in practice
Since 9 July, Salesforce says issues with the package will not be fixed, so a job that starts failing will stay broken. From 31 December the display is read-only, so do not count on editing a configuration after that date. Salesforce’s retirements list calls 31 December the end-of-support date: treat it as the last day, not a target.
Your real deadline is earlier. Salesforce lists a 29-day refresh interval for Full sandboxes, so a Full sandbox may get only one more refresh this year. Any refresh planned for January needs a working Data Mask & Seed policy before it starts, and a holiday change freeze, if you have one, pulls that date into mid-December.
Step 1: Write down what the old package really does
Nothing converts, so the legacy configurations are your only specification. For each saved configuration, record:
- Its name and the sandboxes it runs against.
- Every object and field it masks, with the rule on each. Trailhead lists three legacy rules: random characters or numbers, replacement from a library such as first name or postal code, and deletion.
- The general options: whether case comments are masked and whether all emails and Chatter feeds are deleted. They sit outside the field list and are the easiest thing to lose.
- The last successful run per sandbox, from the Masking Execution Logs tab.
Then look for gaps. Trailhead says the legacy package does not support checkbox, lookup or picklist fields, so a picklist that records something sensitive, such as a vulnerable-customer reason, has never been masked. Confirm how Data Mask & Seed handles those field types in your own org rather than assuming either way.
Step 2: Translate each rule, do not just copy it
Salesforce describes four masking options for Data Mask & Seed: anonymisation, pseudonymisation, pattern matching and deletion. The legacy rules map across roughly like this:
- Random characters or numbers become anonymisation. Carry over any number ranges your tests rely on.
- Library replacement becomes pseudonymisation or a pattern. Decide whether the same customer must get the same replacement on every object.
- Deletion stays deletion, but check each one again: a field deleted in 2023 may now drive a flow you need to test.
Salesforce also says rules can change part of a value or append text: appending “.invalid” to email addresses keeps them readable but undeliverable. If validation rules expect UK postcodes or phone formats, set the policy locale; en_GB is one of the supported values.
Step 3: Keep the new policies as metadata
The legacy configurations lived inside the package. The new policies can live in source control. Salesforce’s DataMaskPolicy metadata type (API version 61.0 and later, 68.0 for locale) stores each policy’s label, description, locale, run frequency and a runOnRefresh flag. Retrieve the policies once they work and deploy them like the rest of your metadata, as in our post on governance-first CI across multiple orgs. Three details catch teams out:
- With runOnRefresh set to true, a policy runs when a sandbox is created or refreshed. It deploys to production without error but never runs there.
- A recurring frequency also needs a scheduled start time, set in the Setup UI, not through the Metadata API.
- The org needs the Salesforce Data Mask licence, and users need the DataMaskAndSeedUserPsl permission set licence before the DataMaskAndSeedUser permission set can be assigned.
Step 4: Run old and new side by side
Salesforce’s notice says the two can be used in the same org at the same time. Refresh two sandboxes from the same production org, mask one with the legacy configuration and the other with the new policy, and check both field by field against your step 1 inventory. A field the old configuration masked but the new one leaves in clear is a regression. Then run your regression tests on the newly masked sandbox. If tests that passed on the legacy output now fail, a new value is breaking a validation or duplicate rule.
Step 5: Retire the package cleanly
Once every live configuration has a tested replacement, Salesforce’s notice says the old package can be uninstalled. First keep the written inventory and any run history you need for audit, update the refresh runbook to name the new policies and remove the legacy permission set assignments.
How we help
Our data migration team can inventory your legacy configurations, rebuild them as policies and test the result in a sandbox before your next refresh. If you also need sandboxes populated with realistic, relationship-intact data, iSyncSF, our Salesforce-native app, handles sandbox seeding and data masking. Whichever tool masks the data, the five steps above still apply.
Frequently asked questions
Do I need to install Data Mask & Seed?
No. Salesforce says it has been a core app since Summer ’26, with no installation. Check the licence and permission set assignments described in step 3.
Can I keep using the old package after 31 December?
Do not plan on it. It has been unsupported since 9 July and its display becomes read-only on 31 December.
Key takeaways
- Nothing converts: write down every legacy configuration, including its general options, before you rebuild it.
- Translate each rule deliberately, check the checkbox, lookup and picklist gaps, and keep the new policies in source control.
- Prove parity by masking two sandboxes side by side and running your tests, then uninstall the old package.

