Test environments and conversion approaches

Test environments and conversion approaches

Published September 28, 2026

Test environments and conversion approaches

On this page

Direct conversion risks operating outages

A London ambulance service abandoned its computer-aided dispatch replacement after implementation failed. With direct conversion, the old system has no live role when the new system is not ready, so the choice can create an operating outage.

Purpose, risks and tooling of change management explains why changes need controlled testing and monitoring. The first comparison identifies the testing level that should provide evidence before conversion begins.

Test levels and their environments

Environment names vary by organization. The test level is identified by scope, purpose, and the person making the decision.

AttributeUnitIntegrationSystemAcceptance
MeaningTests one software component in isolationTests data flow and behavior between connected components or systemsTests the complete integrated solution against its stated requirementsTests whether the customer or business users accept the complete solution against agreed acceptance criteria
Usual environmentDevelopment environment with isolated code and test doubles such as stubs or mocksControlled integration or test environment with connected components and test dataSystem test or staging environment that resembles productionUser acceptance testing environment, normally staging or another controlled production-like environment
ScopeOne function, class, module, or serviceInterfaces, messages, APIs, databases, or connected servicesEnd-to-end application and supporting componentsComplete business workflow and user outcome
Main questionDoes this component work by itself?Do the connected parts exchange and process information correctly?Does the complete solution meet its specified requirements?Can the customer or business owner approve this solution for use?
Typical performerDeveloper or component test teamIntegration test team or developers working across componentsIndependent test team or system test teamCustomer, business owner, or representative users
Evidence or resultComponent test results and defect recordsInterface results, message checks, and integration defectsRequirements-based system test resultsAcceptance results, user feedback, and approval or rejection
Clue words in a scenarioone module, isolated function, developer, stub, mock, componentinterface, API, message, two services, data exchange, connected systemsentire application, end to end, complete solution, system requirementsbusiness users, customer, user acceptance, sign-off, acceptance criteria

The first table connects each test name to its environment, scope, and decision maker.

Change documentation and patching provides the component inventory, baseline configuration, change request, and rollback information. These records identify what is being tested.

AttributeDirectParallelPilot
OperationTurn off the old system and switch all users or sites to the new system at one planned timeRun the old and new systems at the same time for the same operation, then compare or reconcile resultsMove a limited site, department, product, or user group to the new system before wider rollout
Old systemUnavailable after the switch unless restoredContinues to operate while the new system is checkedContinues for the sites or users outside the pilot; it may also run beside the new system at the pilot site
Live disruption riskHighest because the whole operation changes at onceLowest for continuity, although duplicate work and synchronization create cost and control riskLimited to the pilot group, but problems can still affect that group
RollbackDifficult and may require restoring the old system and converted dataUsually easiest because operations can return to the old systemStop or reverse the pilot while the remaining operation stays on the old system
Best clueeveryone switches, old system is shut down, one cutoverboth systems run, compare outputs, duplicate entry, keep the old system availableone branch, one warehouse, limited users, trial site before rollout
Important relationshipDirect is a method of timing the whole changeParallel is a method of keeping both systems operatingA pilot can use either direct or parallel operation within the limited pilot group

The second table changes the question from what is tested to how much of the live operation changes and whether the old system remains available.

Testing change and configuration controls later checks whether these environment and conversion controls were followed. Production is live and does not replace normal unit, integration, or system testing.

Testing and conversion classification rules

For testing, the separating axis is the question being answered. Scope moves from one component, to connected components, to the complete system, and finally to business acceptance.

Users watching a test does not make it acceptance testing. Acceptance requires a business or customer decision against acceptance criteria.

Loading diagram…

For conversion, ask how much of the operation changes at one time and whether the old system keeps running. Direct changes the whole operation at once. Parallel keeps both systems operating. Pilot limits the new system to a defined group.

Pilot and parallel can appear together. One branch may run both systems while the rest of the organization remains on the old system.

The table gives the attributes. The boundary gives the quickest classification rule. The practice cases change the scope or decision maker, then the final case changes rollout coverage.

Separating the four testing levels

Each scenario changes the scope or decision maker, not just the vocabulary. The deciding fact separates the correct test from its sibling tests.

Practice

A developer runs tests on one password-hashing function with mocked database calls before the function is connected to the application. Which type of testing is being performed?

A team connects the new order service to the inventory service and checks whether inventory messages are sent, received, and processed correctly. Which type of testing is being performed?

A test team runs the complete payroll application, database, access controls, and reporting process against the documented system requirements in a production-like environment. Which type of testing is being performed?

Store managers run the final order-to-cash workflows in a user acceptance environment and decide whether the system meets the agreed business acceptance criteria. Which type of testing is being performed?

After these classifications, the conversion approach changes when rollout coverage changes. Test environments and conversion approaches connects the testing level with the later conversion decision.

Changing rollout coverage changes conversion

Was: A company turns off its old order system at midnight and moves every warehouse to the new order system at the same time.

Which one? Identify the conversion approach from these facts.

Answer: The original approach was direct conversion. The whole operation changed at one planned time, and the old system was unavailable.

Now: The company changes the rollout so that only one warehouse uses the new order system first, while the other warehouses continue using the old system.

Which one now? Identify the conversion approach after the fact changes.

Answer: The new approach is pilot conversion. The changed fact is rollout coverage. The new system moved from all warehouses at once to one limited site first, while the old system remained for the other sites.

The old system being unavailable everywhere identifies direct conversion. The limited first site identifies pilot conversion. The takeaways keep these facts separate from the test-level distinctions.

Testing levels and conversion approaches

Unit testing: one component in isolation, usually in development.

Integration testing: connected components or systems exchanging and processing information.

System testing: the complete integrated solution checked against its requirements.

Acceptance testing: business or customer users decide whether the solution meets acceptance criteria.

Direct conversion: the whole operation switches to the new system at one planned time.

Parallel conversion: old and new systems operate together while results or operations are compared.

Pilot conversion: a limited site, group, or department uses the new system before wider rollout.

Testing axis: the scope and decision move from component correctness to business acceptance.

Conversion axis: the amount of the operation changed at once and whether the old system remains available.

Purpose, risks and tooling of change management explains why these facts control change risk. Testing change and configuration controls checks whether the selected environment and conversion controls were actually used.

Unlock the full lesson and continue reading

Signup to continue reading this lesson and unlimited access questions, flashcards, AI notes, and more

Scan to download app

Scan to download
UNLOCK FREE ACCESS

Practice Questions: Test environments and conversion approaches

  1. 1. A developer modified an ERP configuration so that freight charges are allocated among inventory items. The change was approved by IT management, tested by the developer in a development environment, and migrated by a separate release manager. After implementation, accounting discovered that the allocation did not follow its approved policy. Which additional control would have most directly prevented the error?

    • A. Require quarterly recertification of the developer's system access
    • B. Retain prior source-code versions in a controlled repository
    • C. Perform a postimplementation review after the first monthly close
    • D. Require accounting-owner acceptance testing before production migration
    Show answer

    Answer: D. Require accounting-owner acceptance testing before production migration

    Require accounting-owner acceptance testing before production migration - Accounting-owner acceptance testing would check the freight-allocation configuration against accounting’s approved policy before it entered production. - IT management approval, developer testing, and migration by a separate release manager did not establish that the allocation rule met accounting’s requirements. Require quarterly recertification of the developer's system access - Access recertification checks whether the developer should retain system access, not whether the freight-allocation rule follows accounting policy. - The error arose from an incorrect business rule despite IT management approval and a separate release manager. Retain prior source-code versions in a controlled repository - Version retention supports tracing or reversing a change but does not test the new allocation against accounting’s approved policy. - Preserving the prior configuration would not have prevented the incorrect rule from being migrated. Perform a postimplementation review after the first monthly close - A postimplementation review could identify the allocation error, but only after the change had entered production. - Accounting discovered the policy mismatch after implementation, so a review after the first close would not have prevented it.

  2. 2. An ERP database permits each user to hold multiple security roles and each security role to be assigned to multiple users. Which relational design would most appropriately represent these assignments while preserving referential integrity?

    • A. Add RoleID as a foreign key in the User table
    • B. Add UserID as a foreign key in the Role table
    • C. Store each user's assigned RoleIDs in one delimited User-table field
    • D. Create a UserRole table containing UserID and RoleID as foreign keys
    Show answer

    Answer: D. Create a UserRole table containing UserID and RoleID as foreign keys

    Create a UserRole table containing UserID and RoleID as foreign keys - The ERP assignments are many-to-many: each user can have multiple roles, and each role can belong to multiple users. - A UserRole junction table stores one assignment per row, with foreign keys to both tables; the UserID–RoleID pair can serve as a composite primary key to prevent duplicate assignments. Add RoleID as a foreign key in the User table - One RoleID field in each User row permits only one role per user, contrary to the ERP requirement. - Adding another role would require duplicating the user row or changing the design, rather than recording another assignment. Add UserID as a foreign key in the Role table - One UserID field in each Role row permits only one user per role, contrary to the ERP requirement. - Assigning the same role to another user would require duplicating the role row or changing the design. Store each user's assigned RoleIDs in one delimited User-table field - A delimited list puts multiple RoleIDs in one field instead of storing one assignment per row. - Individual RoleIDs in that field cannot be enforced through ordinary foreign-key constraints against the Role table.

  3. 3. After an approved tax-rate change was deployed, the production billing application calculated tax differently from the version that users had tested. Investigation showed that a developer, working under a deadline, bypassed the standard release process and manually rebuilt the program from a local copy immediately before deployment. Which control would best prevent this condition?

    • A. Review developer logical-access privileges every quarter.
    • B. Back up the production application before each release.
    • C. Deploy the approved, version-controlled artifact through a restricted automated pipeline.
    • D. Perform production-response-time testing after each release.
    Show answer

    Answer: C. Deploy the approved, version-controlled artifact through a restricted automated pipeline.

    Deploy the approved, version-controlled artifact through a restricted automated pipeline. - The restricted pipeline deploys the same approved artifact users tested, rather than allowing a developer to rebuild the program from a local copy. - This prevents the untested build that caused production tax calculations to differ from the tested version. Review developer logical-access privileges every quarter. - A quarterly review may identify excessive developer access, but it does not ensure that the tested version is deployed. - The developer could still substitute a local build before the next review. Back up the production application before each release. - A backup supports recovery if a release fails, but it does not stop a local build from being deployed. - The tax-calculation difference would still reach production before any rollback. Perform production-response-time testing after each release. - Response-time testing checks performance, not whether the deployed program matches the approved artifact. - Testing after release also cannot prevent the untested local build from entering production.

  4. 4. An approved tax-calculation change passed user acceptance testing. A production administrator later installed a package with the same file name, but it contained an earlier calculation routine. The change ticket documented approval and testing but did not identify the tested package's version. Which control would most directly prevent this error?

    • A. Require users to repeat acceptance testing after each monthly close
    • B. Use a versioned, signed deployment package and verify its hash before migration
    • C. Review developer access rights to the test environment each quarter
    • D. Reconcile calculated tax expense to tax payments after deployment
    Show answer

    Answer: B. Use a versioned, signed deployment package and verify its hash before migration

    Use a versioned, signed deployment package and verify its hash before migration - A versioned package identifies the calculation routine that passed user acceptance testing, which the change ticket did not do. - Verifying its signature and hash before migration confirms that the production administrator is installing that exact package, not an earlier routine with the same file name. Require users to repeat acceptance testing after each monthly close - Testing after a monthly close occurs after deployment, so it cannot prevent installation of the earlier routine. - Repeating tests also does not establish that the package installed in production is the one users approved. Review developer access rights to the test environment each quarter - Quarterly access reviews address who can use the test environment, not which package the production administrator installs. - The documented approval and testing do not identify the tested version, so this review would not resolve the package mismatch. Reconcile calculated tax expense to tax payments after deployment - A reconciliation performed after deployment might reveal an error but cannot prevent installation of the earlier routine. - Tax expense and tax payments can differ for reasons other than a calculation error, so the reconciliation does not verify the deployed package.

  5. 5. A developer modifies an ERP configuration used to calculate customer discounts. The developer documented unit testing and has no production access. Before implementation, which control would provide the most relevant evidence that the change meets approved business requirements and does not produce unintended accounting results?

    • A. A security administrator's review of the developer's network access logs
    • B. A database administrator's confirmation that the nightly backup completed
    • C. The developer's comparison of the revised program code with the prior version
    • D. User acceptance testing performed and approved by the process owner
    Show answer

    Answer: D. User acceptance testing performed and approved by the process owner

    User acceptance testing performed and approved by the process owner - User acceptance testing checks whether the revised discount configuration meets approved business requirements and produces expected accounting results before implementation. - Process-owner approval provides business-side evidence beyond the developer’s documented unit testing. A security administrator's review of the developer's network access logs - Access logs can help identify unauthorized activity but do not show whether customer discounts are calculated correctly. - The developer has no production access, so reviewing that access does not establish that the change meets business requirements. A database administrator's confirmation that the nightly backup completed - A completed backup supports recovery if data must be restored; it does not test the revised discount calculation. - Backup confirmation therefore provides no evidence that the change produces the accounting results the process owner expects. The developer's comparison of the revised program code with the prior version - A code comparison can identify what changed, but it does not establish that the new configuration meets approved discount requirements. - Because the developer already performed unit testing, another developer-led review does not replace process-owner acceptance of the business and accounting outcomes.

Flashcards: Test environments and conversion approaches

  1. 1. List the risks associated with e-commerce.

    Show answer

    - Data breaches or theft - Reduced paper trail for auditing

Everything you need for CPA prep

Get full Oncourse access with lessons, practice questions, flashcards and AI study tools.

GET STARTED FOR FREE
75% OFFLimited time offer
GET 75% OFF