Skip to content
CodelessOps

/architect · Use cases · Is it any good

The 42 checks

"Is this model any good?"

Who this is for, and the moment

The FD asks whether the model is any good. The honest answer is a list of things to look at, with the evidence for each, and an equally honest list of what could not be checked. Not a score. The 42 tests are that list, run the same way on every model, with Anaplan’s own checklist wording behind each one.

The question

Every test, every model, on the active snapshot.

The result

ScopeAll 42 tests, every model
Tests42six areas
Found something33review candidates
Clear9not found in these exports
Not run0one partial gap, 1.1
AreaTestsFoundClear
1 Performance and size. Where calculation time and cells go, and the formula patterns Anaplan documents as expensive.880
2 Usage and leftovers. What nothing appears to read: candidates to ask about, never verdicts.541
3 Dependencies and change impact. What depends on what, inside a model and between models.550
4 Correctness and maintainability. Totals that may be wrong, repeated logic, and formulas the next builder has to decode.17134
5 Imports, exports and processes. What loads the model, what has stopped running, what sits outside every process.431
6 How far to trust the analysis. Checks the report runs on itself, so every finding says how strong its evidence is.303

Every test

#TestResult
1.1Where calculation effort concentratesfound · 3 Workforce Planning not run: needs the Calculation Effort column
1.2SUM combined with LOOKUP or SELECTfound · 1
1.3Large text line itemsfound · 5
1.4FINDITEM on a large line itemfound · 1
1.5Text joins on a large line itemfound · 2
1.6Per-item functions repeated on every cellfound · 10
1.7Line items with a dimension their formula does not usefound · 12
1.8Where the cells arefound · 4
2.1Modules nothing readsfound · 7
2.2Unread modules that repeat a module in useclear
2.3Calculated line items nothing readsfound · 5
2.4Data loaded but never readfound · 1
2.5Names that say leftoverfound · 4
3.1Line items with the widest change impactfound · 1
3.2Pass-through chainsfound · 4
3.3Circular referencesfound · 1
3.4Which model feeds whichfound · 6 feeds
3.5Dimensions shared across modelsfound · 7
4.1The same calculation under two namesfound · 2
4.2Line items that only copy anotherfound · 14
4.3Formulas that differ in exactly one placeclear
4.4Identical formula text that cannot be comparedclear
4.5The same line item and formula in more than one modelfound · 8
4.6Formulas with more than 10 IFsfound · 1
4.7Hard-coded numbers in formulasfound · 9
4.8Very long formulasfound · 1
4.9SELECT on a fixed period or versionfound · 4
4.10DIVIDE() where a zero divisor shows Infinityfound · 15
4.11Subsidiary views used in calculationfound · 2
4.12Summaries on line items nothing readsfound · 37
4.13Modules with more than 50 line itemsfound · 1
4.14Empty modulesfound · 1
4.15Modules without notesfound · 4
4.16Percentages whose totals are added upclear
4.17The odd one out in a run of matching formulasclear
5.1Imports and exports outside every processfound · 12
5.2Imports and exports with no recent runfound · 7
5.3Imports and exports never runclear
5.4Several imports loading one targetfound · 1
6.1Formulas the parser did not followclear · 0
6.2Agreement with Anaplan's own Referenced Byclear · 0
6.3Input checks on the filesclear · 0

A test that found something is an observation or a review candidate, never a verdict; clear means the pattern was not found in these exports; not run means an export or column the test needs is missing, never that the model is clear.

Evidence drawer · what each chip opens

  • Test6.2 Agreement with Anaplan's own Referenced Byclear
  • Test6.1 Formulas the parser did not followclear
  • Rule42-test catalogue, anaplan-estate v0.2.0checks.py
✓ The checks ran on the active snapshot. Coverage checks 6.1 to 6.3 came back clear: 0 unparsed formulas, 0 disagreements with Anaplan's own Referenced By, 0 input problems.
A real result, re-set for this page. Real app, fictional estate, 12 October 2026.

Six areas, from performance and size through to the three checks the report runs on itself. Thirty-three tests found something to review. Nine came back clear. One partial gap: calculation effort could not be assessed for one model because its export predates the Calculation Effort column, and the app names the owner note that explains why.

The evidence

The catalogue is one numbered registry, published in the Model Health Check and run unchanged inside the Architect. The three coverage checks are the ones to read first: 0 formulas the parser did not follow, 0 disagreements with Anaplan’s own Referenced By, 0 input problems. Below 95% agreement on 6.2 every dependent finding would be downgraded. On this estate it came back clear.

What to notice

  • Found is not failed. The app’s own note under the table: a test that found something is an observation or a review candidate, never a verdict; clear means the pattern was not found in these exports; not run means a column the test needs is missing, never that the model is clear.
  • The biggest counts are housekeeping: 37 summaries on line items nothing reads, 15 DIVIDE() calls that would show Infinity on a zero divisor. Those never compete for the plan. The plan is the next page.
  • Clear is reported with the same prominence as found. Nine tests passed, and the FD gets to see which.

Use cases: The fix list · What nothing reads · What did you test. Technical: Model Health Check.

Run Anaplan in-house?

Start with the Model Health Check on your own exports.

The 42 tests and the action plan run free, hosted or on your own machine, and nothing is stored. Then give me 30 minutes and bring the three questions you would ask about your estate.