GuideBuilding

The export worked. The move didn’t.

Rehearse an actual job in the replacement tool. A complete row count can hide an unusable migration.

AI-generated monochrome photograph of modular metal components moving between two open cases through a fitted transfer fixture.
EXPORT / RECONSTRUCT / RESUMEAI-GENERATED EDITORIAL ARTWORK
THE USEFUL PART

Measure whether the work can continue with its relationships, evidence, and permissions intact—not only whether the files arrived.

A folder of files is not yet a working business

The export finishes. There is a ZIP, a sensible filename, and enough rows to match the dashboard. Then somebody tries to continue the work elsewhere. Attachments point back to the old service. A customer identifier has changed. The approval history is a screenshot. Nobody can tell whether a note was private or visible to the client.

The files arrived. The workflow did not.

Our earlier guide to choosing tools you can leave describes what to ask before adopting a service. This article takes the next step: a small migration exercise that checks whether a real job remains possible after the original interface is unavailable.

Portability is a practical capability, not a format preference. A CSV can be useful and incomplete. An elaborate archive can preserve everything but require more reconstruction than the team can afford. The right test starts with the work you need to resume.

Pick a job with enough history to be awkward

Imagine a small support operation moving twenty-four closed cases into a test workspace. Select ordinary cases and some with the features that create migration risk: several attachments, reassigned owners, corrected details, private notes, a reopened issue, and an approval that changed the next action.

The exercise is not to make the records look similar. Ask another team member to find the correct customer, explain the latest status, open the relevant evidence, identify who approved the decision, and prepare the next response with the correct visibility.

Run this against a separate destination with appropriate access controls. Keep outbound messages, automations, and billing disabled for the drill. A migration test should not resend old customer emails or trigger twenty-four fresh workflows because an import treats every row as new business.

Define the pass conditions before importing. Content, attachments, relationships, timestamps, access rules, and the ability to perform the chosen job are separate checks. Some may be mandatory; some may have an accepted manual workaround. Make that distinction explicit instead of discovering it during the real move.

Count usable cases, not only exported rows

Here is an invented outcome from that twenty-four-case drill. All twenty-four case records are present. Twenty-two have complete usable attachments. Nineteen preserve the relationships needed to interpret the thread. Fourteen have traceable review history. On direct case-by-case inspection, twelve pass every required condition.

Those figures describe overlapping sets, not stages in a funnel. You cannot multiply their percentages to calculate the number of complete cases. The twelve fully usable cases are a separately observed count in this hypothetical example.

The row-count check would report 100% completion. The operational check would report 50%. That gap is the reason to rehearse. It tells you what must be recovered, mapped, rebuilt, or accepted as a limitation before people depend on the destination.

Measure recovery effort too. A missing relationship repaired in thirty seconds is different from an attachment that no longer exists. Keep a log of manual steps and the person who knows how to perform them. A migration that works only while one employee explains every record is still carrying a major dependency.

FIELD NOTES / ROW COUNT IS THE FIRST CHECK
CASE RECORDS EXPORTED24 / 24

100% of records are present

CASES PASSING EVERY REQUIREMENT12 / 24

50% are usable for the selected job

Complete usable attachments22 / 24
Required relationships preserved19 / 24
Review history traceable14 / 24

The checks overlap. Inspect the complete case; do not multiply these percentages.

Hypothetical migration drill. The twelve fully usable cases are a separate case-by-case count, not a value derived from the three marginal completeness checks.

Test the receiving side

The Data Transfer Project provides a framework for transfers between participating services, including shared data models and service-specific components. Its architecture makes a useful point: moving data involves what the source can provide and what the destination can accept. An export format alone does not guarantee that another tool reconstructs its meaning. [1]

Check identifiers and relationships first. Are messages still attached to the right case? Are attachments durable files or links whose access depends on the old account? Do dates preserve their timezone and original meaning? Does an empty field mean unknown, cleared, unavailable, or not applicable?

Then check permissions. A private note becoming a public comment is not a cosmetic mismatch. If the destination has a different access model, map it deliberately and test with an account that has ordinary permissions. An administrator being able to open everything tells you little about what a customer or contractor can see.

For an AI workflow, also carry the instructions, reference sources, acceptance criteria, and tool settings needed to reconstruct the job. A transcript can show what happened without being an executable replacement for the system that made it happen. Test the replacement on representative tasks rather than assuming that moving the prompt recreates the behavior.

Keep export, migration, and deletion separate

An archive is a snapshot taken under particular conditions. Records may change while it is being prepared. Google Takeout's documentation, for example, explains that changes between requesting and creating an archive may not be included, and that downloading data does not delete it from Google's servers. Those are product-specific facts that illustrate two distinctions worth checking in any service. [2]

For a live migration, define how to handle activity after the snapshot: a short write freeze, a supported incremental transfer, or a reconciliation step. Record the cutoff and how missed changes are detected. Do not rely on the ZIP's creation time to explain every record's effective time.

Deletion is another decision. Validate the destination and the retention requirements before removing the source. Keep a rollback path for the period in which errors are still being discovered. An apparently successful import should not automatically trigger irreversible cleanup.

Export permissions may also differ from normal viewing permissions. Confirm that the person carrying out the move is authorized for the records included and that the destination is suitable for them. The drill should improve control over the data, not create an untracked extra copy in somebody's downloads folder.

Rehearse again when the dependency changes

The general recovery discipline is well established: AWS's reliability guidance recommends testing recovery implementations and recording issues found during drills. A software migration is not the same operation as disaster recovery, but the practical lesson transfers: the procedure needs to work when somebody actually follows it. [3]

Repeat a small exercise after meaningful changes to the source schema, destination, access model, or workflow. Keep the test sample representative as new kinds of work appear. Record elapsed time and manual effort, then update the runbook with what actually happened.

A useful exit plan can be modest: a recoverable sample, documented mappings, a destination that has successfully performed the job, and someone responsible for repeating the check. The standard is not that every tool must be interchangeable. It is that when the business decides to move, the work is capable of moving with it.

Sources & notes

Primary references for the product details and research discussed above. The examples and practical recommendations are our editorial interpretation.

  1. Data Transfer Initiative / Data Transfer ProjectData Transfer Project — overview and architecture ↗

    Primary project documentation describes a framework, data models, and service components for transfers between participating providers. It does not guarantee arbitrary app-to-app compatibility.

  2. Google Account HelpHow to download your Google data ↗

    Documents Google Takeout, including archive timing and the distinction between downloading and deleting. These are product-specific examples, not universal export behavior.

  3. AWS Well-Architected FrameworkREL13-BP03 — Test disaster recovery implementation ↗

    Official recovery-testing guidance. Applying the discipline to a software-migration rehearsal is JDGMNT’s editorial adaptation, not a claim that the two operations are identical.

CHECKED / 18 NOVEMBER 2026

About this article

Prepared by JDGMNT Editorial on 18 November 2026. The twenty-four-case migration exercise and its counts are hypothetical. Completeness checks overlap; the twelve fully usable cases are separately observed within the invented example, not calculated by multiplying marginal percentages. References are dated to the source check above. The artwork was generated for this collection.

← BACK TO RESEARCH

JDGMNT / INDEX

Find a useful idea.

JDGMNT / CONTACT

What could be more accessible?

Tell us about a question, an access barrier, or a possible connection to our work.

Your message goes directly to the JDGMNT team. Please don’t include sensitive personal information.

EXPLORE JDGMNT

MORE KNOWLEDGE. MORE POSSIBILITY.
MAKING ACCESS POSSIBLE.