Who this is for, and the moment
Before you trust a report you ask what it did not do. The chat answers that from the test catalogue, with the numbers, the tests that came back clear, the one gap and the owner note that explains it, and the list of what the exports cannot hold at all.
The question, as typed
“What did you test on this estate? What came back clear, and was there anything you could not run?”
The answer
Brief
The report ran 42 tests across Caldergate Data Hub, Caldergate FP&A, Workforce Planning and Board Reporting: 33 found something to review, 9 came back clear, and 0 were wholly not run. Clear means the pattern was not found in these exports, not proof it cannot exist. The only partial gap is test 1.1, where calculation effort could not be assessed for Workforce Planning because the export lacks the Calculation Effort column, which matches owner note n-ff27724850. The strongest coverage checks came back clear: test 6.1 found 0 unparsed formulas, test 6.2 found 0 Referenced By agreement issues, and test 6.3 found 0 input guard issues.
What was tested?
| Area | Tests run | Found | Clear | Not run |
|---|---|---|---|---|
| 1 performance and size | 8 | 8 | 0 | 0 |
| 2 usage and leftovers | 5 | 4 | 1 | 0 |
| 3 dependencies and change impact | 5 | 5 | 0 | 0 |
| 4 correctness and maintainability | 17 | 13 | 4 | 0 |
| 5 imports, exports and processes | 4 | 3 | 1 | 0 |
| 6 how far to trust the analysis | 3 | 0 | 3 | 0 |
| Total | 42 | 33 | 9 | 0 |
Which tests found something?
The 33 tests that found something are review candidates, not automatic defects. The largest: 4.12 summaries on line items nothing reads, 37 hits; 4.10 DIVIDE() where a zero divisor shows Infinity, 15; 4.2 line items that only copy another, 14; 1.7 line items with a dimension their formula does not use, 12; 5.1 imports and exports outside every process, 12.
What came back clear?
| Test | Title | Result |
|---|---|---|
| 2.2 | Unread modules that repeat a module in use | 0 hits |
| 4.3 | Formulas that differ in exactly one place | 0 hits |
| 4.4 | Identical formula text that cannot be compared | 0 hits |
| 4.16 | Percentages whose totals are added up | 0 hits |
| 4.17 | The odd one out in a run of matching formulas | 0 hits |
| 5.3 | Imports and exports never run | 0 hits |
| 6.1 | Formulas the parser did not follow | 0 hits |
| 6.2 | Agreement with Anaplan's own Referenced By | 0 hits |
| 6.3 | Input checks on the files | 0 hits |
What could not be run?
No test was wholly not run: the test catalogue reports 0 not-run tests out of 42.
There is one partial coverage gap. Test 1.1, where calculation effort concentrates, could not assess Workforce Planning because it needs the Calculation Effort column. Owner note n-ff27724850 says the Workforce Planning export was taken before Calculation Effort existed, so ask for a fresh export before judging where calculation time goes in Workforce Planning.
What is outside the exports?
The exports do not show saved views and filters, UX pages, dashboards, selective access, list members and subsets, planning data values, model history, or who changed what. That means these tests can review structure, formulas, dependencies, actions and metadata, but they cannot prove whether a reported number is correct or whether a line item is used only by a page or saved view.
Evidence drawer · what each chip opens
- ModelCaldergate Data Hublist_modelsobserved
- ModelCaldergate FP&Alist_modelsobserved
- ModelWorkforce Planninglist_modelsobserved
- ModelBoard Reportinglist_modelsobserved
- Owner noteWorkforce Planning: naming and the missing effort columnnotesuser confirmed
Forty-two tests, 33 found something, 9 clear, 0 not run. The nine clear tests listed. The one partial gap, with the note that says the export was taken before the Calculation Effort column existed. And a plain paragraph on what is outside the exports.
The evidence
Five citations: the four models and one owner note. Sixty-one figures in the answer, all 61 traced to a tool result. Nothing named without being looked up. That is the cleanest verdict line on this site, and it is on the answer about the app’s own limits.
What to notice
- The architect reports its own gaps. “Clear means the pattern was not found in these exports, not proof it cannot exist” is in the answer, not in a footnote I added.
- The chat reads the same catalogue the Tests page shows. There is one registry of checks; the answer cannot drift from the page.
- Every session is a transcript kept with the estate. The question a new joiner asks in March is still there in September, with its answer and its citations.
Related
Use cases: The 42 checks · The estate on day one. Technical: Engineering record.