Lesson 4 of 5
Data hygiene
9 min
Listen to this lesson
Every analytics view in this course rests on people keeping Cloud ALM data accurate: linking items, updating statuses, finishing test runs, closing defects properly. When that slips, the dashboards still produce numbers, just wrong ones. Data hygiene is the unglamorous work that makes a gate honest.
The usual problems
| Problem | What it does to the gate | How to find it |
|---|---|---|
| Test cases not linked to requirements | Coverage looks lower than reality, or tests do not count toward any requirement | Filter test cases without requirement or user story assignment |
| Requirements not assigned to scope or process | They drop out of process-level views | Filter requirements without scope or solution process |
| Test runs started but never finished | Execution figures stall; results are ambiguous | Test execution list filtered by in-progress runs older than a few days |
| Stale statuses | Requirements or user stories still "in progress" though done, or "done" though untested | Items not changed for weeks compared with their linked test status |
| Defects without priority, owner or scope | They vanish from priority and process views | Defect list filtered by empty fields |
| Duplicate test cases or defects | Inflated counts, split evidence | Search by title and process; merge or close duplicates with a note |
Common Cloud ALM data problems and their effect on gate reporting.
A weekly hygiene routine
- Run the hygiene filtersSaved filters for each problem in the table above
- Assign fixesEach item to its owner: process test lead, triage lead or requirement owner
- Fix at the sourceAdd the link, finish the run, set the priority; do not adjust the report
- Re-run the filtersConfirm the counts dropped
- Report hygiene scoreOpen hygiene issues by type, tracked week by week
Make hygiene a standard, not a favor. Write the rules into the test strategy: every test case must link to at least one requirement or user story; every defect needs priority, owner and scope at creation; test runs are finished or explicitly aborted at the end of each day. Use Cloud ALM tags where your teams need sub-classifications the standard status model does not cover, but agree the tag list centrally so it stays usable for filtering.
Who owns hygiene
- Test lead: defines the rules and the filters, reports the hygiene score.
- Process test leads: fix links and test statuses in their process.
- Triage lead: fixes defect fields and duplicates during daily triage.
- Requirement owners: keep requirement status and scope assignment current.
- Cloud ALM administrator: maintains roles, saved views and any tag lists.
Hygiene starts with set-up
Many hygiene problems are created on day one. Agree the structure before the first test case is written: how scope and solution processes are named, which requirement and user story fields are mandatory, how test plans are cut per cycle (for example one plan for SIT cycle 1, one for SIT cycle 2, one for UAT), and which roles may change statuses. A test plan per cycle matters for gates, because it lets you report execution for exactly the tests in scope for that gate without filtering out last cycle's runs. Fixing structure halfway through a program is possible, but it is far more expensive than setting it up well.
- Before the first test caseNaming agreedHow scopes and solution processes are named
- Before the first test caseMandatory fields setWhich requirement, user story and defect fields must be filled
- Before each cyclePlan per cycleOne plan for SIT cycle 1, one for SIT cycle 2, one for UAT
- Before each cycleStatus rights fixedWhich roles may change test case and requirement statuses
- Every weekHygiene routine runsSaved filters checked and fixes assigned to data owners
- Does a gate figure look wrong?Yes → find out why before the gate · No → snapshot it with filter, date and author
- Is the cause missing links, unfinished runs or stale statuses?Yes → the data owner fixes it in Cloud ALM, then re-run the view · No → next question
- Is the figure right but unwelcome?Yes → report it as it is, with a risk statement; never adjust it in a slide or spreadsheet · No → snapshot and report
What would you do?
A week before the gate, Contoso's order-to-cash view shows 15 test runs still "in progress" from ten days ago. The testers say they finished the tests but forgot to complete the runs. What should Priya do?
- Unlinked items, unfinished runs, stale statuses and empty defect fields quietly distort gate reports.
- Run saved hygiene filters weekly and assign fixes to the owners of the data.
- Write hygiene rules into the test strategy and track a hygiene score.
- Always fix data at the source in Cloud ALM, never in the slide.
Check your understanding · not graded
Lesson check
1. What happens to coverage when test cases are not linked to requirements?
2. Where should a wrong gate figure caused by missing links be corrected?
3. Which defect fields should be mandatory at creation for gate reporting?
4. Why agree a central list of tags?