Lesson 2 of 4 · ERP Programme Testing

Oracle Fusion Cloud Testing

Oracle Fusion is the dominant ERP in NZ government (AoG panel). Learn to navigate its module architecture, test finance and HR workflows, and understand the three layers every change requires testing at.

ERP Testing ERP Programme Testing — Lesson 2 of 4 ~30 min read · ~70 min with exercises

1 The Hook

A large NZ government agency ran Oracle Fusion Financials. A new cost centre structure was activated mid-financial year — approved by the CFO, implemented by the Fusion team overnight. No testing was done because “it’s just a configuration change.”

The problem: the new cost centre hierarchy changed how the budget transfer rules evaluated department codes. Budget transfers that had previously routed and posted correctly now failed silently — they appeared to succeed in the UI but the underlying FBDI (File-Based Data Import) that fed the nightly batch did not recognise the new codes. Six months of management reports showed incorrect spend data by department. The budget holders had been making decisions on wrong numbers.

The testers had only tested the UI change — “the new cost centre appears in the dropdown.” Nobody had tested the FBDI import that consumed cost centre codes downstream, or the OTBI reports that displayed the results. That is the lesson: every Oracle Fusion change has three layers, and all three must be tested.

2 The Rule

Every Oracle Fusion change has three layers to test: (1) the UI workflow and approval routing, (2) the underlying flex fields, business rules, and configuration, and (3) the integration interfaces — FBDI imports, REST APIs, and ESS scheduled jobs. Miss any layer and you have a gap that will surface in production.

3 The Analogy

Analogy

Oracle Fusion is like an iceberg.

The UI is 20% above the water — visible, clickable, demonstrable in demos. Below the surface: ESS (Enterprise Scheduler Service) jobs that run nightly, FBDI file imports that load bulk data, OTBI reports that aggregate and display results, REST APIs that talk to payroll and HR systems, and the business rule engine that enforces delegation limits, account codes, and approval hierarchies.

Most testers test the tip. ERP-qualified testers test the iceberg. A UAT sign-off based only on UI walkthroughs is a sign-off on 20% of the system.

4 Watch Me Do It

Here is a complete test case for Oracle Fusion Financials — testing a journal entry through the full approval and posting cycle, including the OTBI report output:

Module: Oracle Fusion Financials → General Ledger
Scenario: Manual journal entry — approval routing and posting verification

Steps:
1. Navigate to Journals → Create Journal
2. Enter: Ledger = NZ Primary Ledger, Currency = NZD, Category = Manual
3. Enter debit line: Account 1234.000.0001 (Dept Cost Centre) $5,000
4. Enter credit line: Account 2345.000.0000 (Control Account) $5,000
5. Verify balance = $0.00 before submit (journal must balance)
6. Submit for approval → verify workflow routes to correct approver
(check approval rule: amount > $2,000 = Finance Manager required)
7. Log in as Finance Manager → verify journal in worklist
8. Approve → verify status changes to “Posted”
9. Run OTBI Trial Balance report for the period → verify both lines appear
10. Verify account codes in Trial Balance match the entered values exactly

Expected: Journal balances, routes to correct approver, posts, appears in OTBI.
Failure modes to check: unbalanced journal rejected; wrong approver routed;
posted journal not in OTBI; wrong account code in report output.
Pro tip: Step 10 is the one most test cases omit. The journal posting to “Posted” status does not guarantee the OTBI report shows the correct account code. OTBI uses a separate reporting layer. Always verify the report output, not just the transaction status.

5 Decision Tool

Oracle Fusion offers multiple testing tools. Choosing the right one depends on what you are testing and the project stage:

ToolBest forLimitations
Oracle Test Manager (OTM)Structured test case management, linking test cases to requirements, defect tracking within the Oracle ecosystemRequires Oracle licencing; limited to test management, not execution automation
Selenium / PlaywrightUI regression automation for stable Fusion workflows; end-to-end smoke tests after quarterly updatesOracle Fusion UI is dynamic; selectors break on quarterly cloud updates; maintenance overhead is high
Manual testingConfiguration testing, FBDI validation, first-pass SIT, UAT facilitation, exploratory testing around new configurationNot scalable for regression; cannot cover FBDI batch processing volumes automatically
FBDI template testingBulk data load validation — upload a test file, verify ESS job output, check error reportsRequires correct file format; column order is version-sensitive; test with realistic row counts

6 Common Mistakes

🚫 Not testing FBDI file format headers after a quarterly Oracle update

Why it happens: FBDI templates look the same across releases and teams assume they are stable.
The fix: Oracle occasionally changes column order or adds mandatory columns in quarterly cloud updates. Always download the current FBDI template from the Oracle Help Centre after every update, compare it to the one your integration uses, and run a test load before the update goes to production. A column order mismatch causes silent data corruption, not an obvious error.

🚫 Testing only the happy path through an approval workflow

Why it happens: The approval path is what the demo showed, and it passes, so the tester moves on.
The fix: Approval workflows have multiple paths: approve, reject, send back for correction, escalate on timeout, and withdraw. Test all of them. A rejection path that fails means approvers cannot send work back to submitters, which blocks the entire process in production. Timeouts and escalations are commonly untested and commonly broken.

🚫 Forgetting to test ESS scheduled jobs that run overnight

Why it happens: ESS jobs are invisible during the business day and not part of the UI walkthrough.
The fix: Many Oracle Fusion processes depend on ESS jobs: GL posting, period close sweeps, intercompany netting, and report generation. Run these jobs manually in the test environment and verify their output. If an ESS job fails silently in production, the consequences appear hours later when users find stale data or unposted transactions — hard to diagnose, expensive to fix.

7 Now You Try

🤖 Prompt Lab — Try it live

Paste or edit the prompt below, then click Run to get a real AI response.

8 Self-Check

Click each question to reveal the answer.

Q1: What are the three layers of Oracle Fusion that must be tested on every change?

1. The UI workflow and approval routing — what users see and interact with. 2. The underlying flex fields, business rules, and configuration — the cost centre codes, delegation limits, account rules that drive the process. 3. The integration interfaces — FBDI file imports, REST APIs connecting to external systems, and ESS scheduled jobs that run the overnight processing. A test suite that only covers the UI is covering roughly 20% of the system.

Q2: What is FBDI and why does it matter to Oracle Fusion testers?

FBDI stands for File-Based Data Import. It is Oracle Fusion’s bulk data loading mechanism — a set of standardised spreadsheet templates that the system imports via an ESS job. FBDI matters to testers because it is how most integration data enters Oracle Fusion (general ledger journals, supplier invoices, asset additions), and it is version-sensitive: Oracle can change column order or add mandatory columns in quarterly updates. A tester who does not validate FBDI files will miss the entire class of bulk-data defects that only appear when large datasets are processed.

Q3: How should you approach regression testing when Oracle does quarterly cloud updates?

Quarterly updates are mandatory in Oracle Fusion Cloud — you cannot opt out. The approach is: (1) maintain a core regression pack of smoke tests covering the highest-risk business processes; (2) run this pack in the test environment when Oracle previews the update, before it reaches production; (3) download updated FBDI templates and compare them to your current integration templates; (4) check Oracle’s release notes for any changes in the modules you use; (5) prioritise testing any workflow or configuration that Oracle has noted as changed. Selenium/Playwright automation helps but requires maintenance after each update because Oracle’s UI can shift.

9 Interview Prep

Real questions asked in NZ QA interviews for Oracle Fusion roles. Read the model answers, then practise your own version.

“How do you approach regression testing when Oracle does quarterly cloud updates?”

Oracle Fusion Cloud updates quarterly and you cannot opt out, so regression is a recurring programme activity, not a one-off. I maintain a regression pack of critical business process smoke tests — typically 30 to 50 test cases covering the highest-risk workflows. When Oracle previews an update I run this pack in the preview environment before the update reaches production. I also check the release notes specifically for changes to modules we use, download updated FBDI templates and diff them against the ones our integrations use, and test any workflow Oracle has flagged as changed. The key is making regression a planned event with a defined scope, not a panic after go-live.

“What is FBDI and why does it matter to testers?”

FBDI is Oracle Fusion’s File-Based Data Import mechanism — spreadsheet templates that load bulk data via ESS jobs. It matters to testers for three reasons. First, it is how most integration data enters Oracle, so if the FBDI is broken, nothing downstream works correctly. Second, it is version-sensitive — Oracle changes column layouts between releases, which can break an integration silently if the integration team does not notice. Third, FBDI errors often appear in an error report rather than as an on-screen failure, so a tester who only checks the UI will miss them. I always include an FBDI test in any integration test plan: upload a test file, trigger the ESS job, check the output and the error report, and verify the imported data looks correct in the application.

Senior engineer insight

Oracle Fusion has thousands of configuration parameters, and the combinations that matter are the ones your organisation actually uses — not the ones Oracle's demo environment uses. Before writing a single test case, spend time with the configuration workbook and identify which intersections of module settings, business rules, and data types apply to your deployment. That is your actual test scope. Everything else is optional. The most expensive Oracle Fusion bugs are not in the UI — they are in the calculation engine, and they only appear with specific combinations of data that no one tested because no one thought to ask what data combinations are actually possible.

From the field

An Auckland logistics company migrated to Oracle Fusion Financials and ran a six-week testing programme before go-live. UAT passed with 94% coverage. Three days after go-live, accounts payable discovered that supplier invoices with a specific combination of payment terms (net-60 with early payment discount) were being processed at the wrong discount rate. The issue had been present in UAT data but the test data never used that combination. Oracle Fusion has hundreds of configuration intersections in financial modules — "testing the happy path" is not sufficient coverage. Since then, the team uses a configuration matrix to generate test cases from the Cartesian product of payment terms, supplier types, and invoice categories. ERP testing coverage is combinatorial, not sequential.