The new system contains the customer's name and current contract amount. It does not show why the amount changed, which version the customer approved, or which invoice your team still needs to correct.
A row count can pass while operational history fails. Scope the migration around the records people must understand and act on after the move.
Separate current work from archive-only material
In this hypothetical migration, a service business is moving its project records into a new application. Active projects need usable status, decisions, attachments, and open exceptions. Closed projects need enough history for staff to answer customer questions and retrieve agreed documents.
Decide what belongs in the working application and what can remain in a searchable archive. Keeping an archive may avoid rebuilding old features, but staff need a supported way to find the record and interpret it. A folder of unidentified exports will not provide that continuity.
Write acceptance cases that connect the records
| Example record | What must survive | Acceptance check |
|---|---|---|
| Project P-104, approved change | Original amount, revised amount, approving person, approved document version | Reviewer traces the revised amount to the specific approval |
| Project P-118, open exception | Unresolved invoice correction, responsible person, next action | Correction appears in the new team's working queue |
| Closed project P-076 | Completion date, customer files, final accepted scope | Support retrieves the accepted document from the archive |
| Customer with two projects | Correct customer-to-project relationships | No project or attachment appears under the wrong customer |
These invented records demonstrate the acceptance pattern. Choose your actual cases before the transfer so the team does not select only records that happened to move cleanly.
Resolve ambiguous source records before interpreting them
The old system may use “complete” for both delivered work and cancelled work. Ask the business to decide how each should appear in the new system. A migration script cannot settle that meaning without a business rule.
Keep the original value when it supplies useful context. If you map several old states into one new state, document that mapping and the information you will lose. Decide whether a history field or archive reference preserves what staff need.
Also check time zones, original identifiers, and file references where those details affect understanding. A date shifted across midnight can change a service commitment. A replacement identifier can break the connection to an external invoice unless the team retains a mapping.
Plan the work that changes during the move
Staff may update records while the migration runs. Agree on the transfer window, which system accepts changes, and how the team captures updates made after the initial export.
For the illustrative business, the first copy happens in a test environment. Staff inspect the chosen acceptance cases there. At cutover, the team uses a defined update window and reconciles the records changed since the first copy before opening the new application for normal work.
Have a recovery choice ready if a material case fails. That might mean postponing cutover or keeping a defined set of work in the old system. State who can make that call and how staff will know where to continue.
Accept a usable history, not just a successful import
Count transferred records and files, but also have staff perform the tasks the history supports: explain a price change, finish an open exception, and locate the customer-approved version. Those checks test whether the transfer preserved operational meaning.
Our guide to one system or connected tools covers whether to move in the first place. For a defined migration and acceptance plan, explore custom software development with The Ops Guide.

