Who this is for, and the moment
Week one on an estate somebody else built. Four models, a shared drive of exports, and one person who knows how it fits together, in meetings until Thursday. The usual route is three months of clicking through modules. This page is day one with the clicking already done.
The question, as typed
“Explain this estate.”
The answer
Brief
This is a four-model Anaplan estate with Caldergate Data Hub, Caldergate FP&A, Workforce Planning and Board Reporting. The estate holds 144,543,801 cells in the exported metadata, calculated as 8,060,698 in Caldergate Data Hub + 130,001,662 in Caldergate FP&A + 6,343,559 in Workforce Planning + 137,882 in Board Reporting; Caldergate FP&A is 89.94% of that. The operating flow is hub to planning, planning to reporting: Caldergate Data Hub feeds Caldergate FP&A and Workforce Planning; Workforce Planning feeds headcount cost into Caldergate FP&A; Caldergate FP&A and Workforce Planning feed Board Reporting. All live-looking model feeds are still inferred from import action names, except Caldergate FP&A to Workforce Planning, which is retired because owner note n-ff27724850 says HR analytics now keys the assumptions by hand. The main handover risks are the stopped FX file, duplicate Workday loads, and Budget v2 DO NOT USE still being selected by a board pack line, from owner note n-07669fb656.
How the models fit together
| From | To | State | Imports | Targets |
|---|---|---|---|---|
Caldergate Data Hub | Caldergate FP&A | inferred | 5 | DAT01 Actuals GL, DAT02 Actuals Volumes, SYS02 Cost Centre Attributes, SYS04 Product Attributes, SYS03 Account Attributes |
Caldergate Data Hub | Workforce Planning | inferred | 1 | Data - Cost Centre Map |
Workforce Planning | Caldergate FP&A | inferred | 1 | INP03 Headcount |
Caldergate FP&A | Board Reporting | inferred | 2 | DAT01 P&L, DAT02 Board Lines |
Workforce Planning | Board Reporting | inferred | 1 | DAT03 Headcount |
Caldergate FP&A | Workforce Planning | retired | 1 | Inputs - Settings |
All arrows are dashed because the feeds are inferred from import action names, or retired. There are no owner-confirmed live feeds in list_models. The retired feed is important. Caldergate FP&A still has an import-name relationship into Workforce Planning, but owner note n-ff27724850 says Import from Caldergate FP&A - Assumptions into Inputs - Settings has not run since August 2022 and HR analytics key the rates by hand.
What each model does
| Model | Role | Scale |
|---|---|---|
Caldergate Data Hub | Loads NetSuite, Salesforce and Workday, then feeds the spokes. Confirmed in owner note n-5f84264ff2. | 12 modules, 63 line items, 8,060,698 cells, 8 imports |
Caldergate FP&A | Plans revenue, opex, headcount cost and cash, and produces management and board packs. Confirmed in owner note n-5f84264ff2. | 30 modules, 278 line items, 130,001,662 cells, 10 imports |
Workforce Planning | Holds employee-level cost. Data - Employees is loaded from Workday, Calcs - Cost inflates salary and adds NI and pension, and Calcs - Headcount by CC is what FP&A imports into INP03 Headcount. Confirmed in owner note n-ff27724850. | 9 modules, 55 line items, 6,343,559 cells, 3 imports |
Board Reporting | KPI layer for the board dashboard, refreshed from FP&A and Workforce after each close. Confirmed in owner note n-5f84264ff2. | 7 modules, 30 line items, 137,882 cells, 3 imports |
Where the size sits
| Model | Cells | Share of estate |
|---|---|---|
Caldergate FP&A | 130,001,662 | 89.94% |
Caldergate Data Hub | 8,060,698 | 5.58% |
Workforce Planning | 6,343,559 | 4.39% |
Board Reporting | 137,882 | 0.10% |
That means most size work should start in Caldergate FP&A, not in Board Reporting.
What to know before changing it
Ownership is split. Caldergate FP&A and Caldergate Data Hub sit with Finance Systems, Workforce Planning belongs to HR analytics, and Board Reporting is Dan’s model. Owner note n-cdfac253fb says any import change must be agreed with the receiving model’s owner and made in a development copy first.
Build history matters. Owner note n-6029978175 says Caldergate Data Hub and Caldergate FP&A were built in 2019 by the first partner using DISCO naming, Workforce Planning was built in 2021 by a second partner using its own naming, and Board Reporting was built in-house by Dan in 2023. The conventions do not agree, and nobody has renamed objects because saved views may break.
Do not treat every long formula as bad. Owner note n-d1f2fe4d49 says CAL10 Depreciation.Charge is correct, reconciles to the fixed asset register, and should not be touched without the asset schedule.
Main handover risks
Owner note n-07669fb656 flags three risks that are not ticketed:
- The FX file has not been loaded since November 2023, so FP&A is running on a hardcoded rate.
- Workday is loaded twice, once by
Caldergate Data Huband once byWorkforce Planning, so the employee lists can drift. Budget v2 DO NOT USEis still selected by a board pack line.
I would treat those as control risks before doing cosmetic cleanup.
What the export does not prove
The export does not show saved views, UX pages, dashboards, selective access, list members and subsets, planning data, model history, or whether a number is correct. It shows structure, formulas, actions, dependencies and parsed metadata. So this is enough to explain the estate and prioritise investigation, but not enough to certify the output values.
Evidence drawer · what each chip opens
- ModelCaldergate Data Hublist_modelsobserved
- ModelCaldergate FP&Alist_modelsobserved
- ModelWorkforce Planninglist_modelsobserved
- ModelBoard Reportinglist_modelsobserved
- Owner noteWhat the estate is fornotesuser confirmed
- Owner noteHandover risks nobody has ticketednotesnot established
- Owner noteWho built what, and whennotesuser confirmed
- Owner noteOwners, and the rule for changing a feednotesuser confirmed
- Owner noteWorkforce Planning: naming and the missing effort columnnotesuser confirmed
- Owner noteDepreciation Charge: correct, long, do not touch without the schedulenotesuser confirmed
Four models, the operating flow in one sentence, every feed between them with its state, the scale of each model, where the size sits, who owns what, how it was built, and the three handover risks. Every model name is a chip that opens the model as the export holds it; every owner note is cited by id.
The evidence
Ten citations: the four models, handed by the list_models tool, and six owner notes. Sixteen names in the answer were never looked up this conversation (modules, import actions, one line item), and the app says so under the answer rather than passing them off as checked. Twenty-seven figures, every one traced to a tool result or derived from ones that were.
What to notice
- The feeds are marked inferred. The exports carry import action names, not a model-to-model mapping, so the app says “inferred from import names” until an owner confirms the feed. One is marked retired, with the note that retired it.
- The risks come from a note, not from the exports. The FX file that stopped in November 2023 is invisible to any test; a colleague wrote it down, and the answer carries it on day one.
- “What the export does not prove” is part of the answer. Saved views, pages, dashboards and model history are not in the exports, and the app will not pretend otherwise.
Related
Use cases: Why it was built this way · What nothing reads · Why the model is this size.

