When someone asks what to check before a portal migration, I start with the things that are easy to miss. Custom properties and records are visible. A workflow dependency, a permission, or an external integration can be less obvious until it stops working.

This is the checklist I use to structure that conversation. The exact transfer method depends on the source, the target setup, and the platform capabilities available to the project.

1. Inventory the whole process

List the objects and records, then work outward. Identify what creates them, what updates them, and what relies on them.

  • Custom properties, object relationships, and lifecycle rules
  • Workflows, forms, lists, and reporting dependencies
  • Permissions and access requirements
  • Subscriptions, email assets, and the rules governing their use
  • Connected apps, webhooks, custom code, and external services

2. Decide what should move

A migration is a useful point to review duplicates, unused properties, and workflows that no longer serve a purpose. Agree on those decisions with the people responsible for the data before changing or excluding anything.

Keep a record of what will move, what will be transformed, and what will be left behind. That record becomes part of the validation plan.

3. Map meaning and relationships

Check property types, internal names, required values, and association rules. Two fields with similar labels can have different meanings. Confirm those meanings with the business.

For an illustrative example, imagine an old system with separate customer and billing-contact records. A migration plan needs to explain how those records map to the target objects, how the relationship is preserved, and how you will check it afterward.

A field mapping should include the source, destination, transformation rule, and treatment of missing or invalid values.

4. Rehearse the difficult cases

Use a suitable test environment and representative sample. Include incomplete records, duplicates, unusual relationships, and data that needs transformation.

Check the behavior of automation around the migration. Establish which processes should run, which should be paused, and how they will be tested and restored. Account for permissions and integrations that need reconfiguration or reauthorization.

Turn the sample into acceptance checks

For the customer and billing-contact example, write down the expected result before the test import. These are illustrative records and rules, not results from a customer project.

Illustrative migration acceptance checks
Source caseExpected resultHow to check
Customer C-104 with billing contact P-22One company and one contact, linked using the agreed billing-contact relationship.Resolve both source IDs in the target and inspect the relationship.
Customer C-105 has no billing contactCreate the company. Record the missing contact for review without inventing a relationship.Confirm the company exists and the exception identifies C-105.
Two customer rows share source ID C-106Hold the conflicting rows for review before creating a company.Confirm neither row created a company and the exception contains both source references.

5. Agree on the checks before cutover

Compare the expected and actual results. Alongside record counts, inspect relationships, field values, exceptions, and the workflows people use. Ask a sample of users to try the process.

Decide how to handle changes made during the migration, who approves the result, and what happens if an important check fails. Prepare the recovery approach before switching systems.

The useful question at the end is whether the team can trust the data and carry out its work in the new setup.

If two source fields appear to mean the same thing, settle their meaning and ownership before finalizing the mapping.

Before you sync that field.