Lesson 5 of 5
Documenting and refreshing data
9 min
A scenario is only rerunnable if its data can be found or rebuilt. Test systems are refreshed, data is consumed, people leave. The answer is a simple, disciplined record of which data each scenario uses and how to recreate it.
The test data sheet
Keep one data sheet per scenario (or one shared workbook with a tab per process). It lists every record the scenario needs, where it came from, and its state before the run.
| Field | Example for O2C-010 |
|---|---|
| Scenario and data set | O2C-010, set 2 of 3 (Maria) |
| System and client | QAS, client 100 |
| Business partner | Fleet test customer tagged UAT-O2C-MARIA-02, customer role in sales area 1710 / 10 / 00 |
| Product and location | CMC-ALT-100, plant 1710, storage location 171A |
| Price data | Contract price condition 240.00 USD per unit, valid from period start to year end |
| Preconditions before run | 200 units unrestricted; credit limit 50,000 USD, exposure 0 |
| How created | Business partners by mass upload template v3; stock by movement type 561 |
| Consumed by run | 50 units stock; 12,000 USD exposure |
| Documents created | Order, delivery, goods issue, billing numbers recorded at run time |
Data sheet fields with a Contoso example.
- PlanList the records the scenario needs and assign a tagged data set
- CreateBuild the records with templates and setup postings
- Capture before stateRecord stock, exposure and open items before step 1
- Run and consumeExecute the scenario; note what the run used up
- RecordWrite document numbers into the data sheet as they are created
- Retire or resetMark the set as consumed, or rebuild it for a retest
Surviving a system refresh
A refresh copies one system (often production, or a "golden" test client) over another. Everything testers built in the target is gone: business partners, stock, open items, test documents. If the data was created by hand and not documented, the next cycle starts with days of rework. If it was created by a documented, repeatable method, it can be rebuilt in hours.
- Before refreshSnapshotExport the data sheets and note which data sets are still needed
- Refresh dayCopy and maskBasis refreshes QAS client 100 and applies the approved masking
- Refresh + 1 dayRebuildRun the mass upload templates and setup postings from the data sheets
- Refresh + 1 dayVerifySmoke test one scenario per process using the rebuilt data
- Refresh + 2 daysReleasePublish updated data sheets to testers and resume the cycle
Making creation repeatable
- Templates over clicks. Keep mass upload templates for business partners, products and condition records under version control.
- Ordered setup steps. Write the precondition postings (stock entry, open items, credit limits) as a short setup script with the same care as a test.
- Naming and tagging rules. Agree tags and number ranges once and reuse them every cycle.
- Ownership. Name who rebuilds each data area after a refresh: Maria for order to cash, the procurement key user for procure to pay.
Retiring data
Data sets have an end. When a scenario has passed and its evidence is filed, mark its data set as consumed in the data sheet so nobody reuses it by accident. Records that must not be used again, such as a test customer deliberately left credit blocked, can be flagged or blocked so they cannot slip into another scenario. At the end of the project, the data sheets and templates move to the regression library: the same scenarios will be run after every upgrade, and the data plan is half the work.
What would you do?
Luis announces that QAS will be refreshed from production in two weeks for a cutover rehearsal. Priya's team is mid-way through UAT cycle 2. What should Priya do first?
- Keep a data sheet per scenario: records, source, before state, consumption, documents created.
- Refreshes wipe test data; documented templates and setup scripts make the rebuild fast.
- Smoke test after a rebuild before releasing the client to testers.
- Name an owner for rebuilding each data area.
Check your understanding · not graded
Lesson check
1. What happens to test data in a client when it is refreshed from production?
2. Which item belongs on a test data sheet?
3. What makes test data quick to rebuild after a refresh?
4. When should document numbers be written into the data sheet?