Academy of Profit Center — Your Journey
Foundations Hall
Hall 1 of 9 · Academy of Profit Center · EPCM Study Tour
The $134M blind spot
NovaPrism Group's CFO cannot tell the board which divisions make money. Not because the data is wrong — because the costs have never been properly allocated. This hall is your first day at the Academy: why this problem exists, how EPCM solves it, and what you must know before entering any other hall.
The profitability blind spot
What NovaPrism's P&L looks like before allocation — and what it looks like after.
NovaPrism Group runs three revenue-generating divisions and one internal cost centre. On paper, Infrastructure Solutions is the largest division by revenue at $980M and appears to earn an 11.2% margin. The finance team reports this to the board. Capital allocation decisions are made on this basis.
The problem: Enterprise Support costs $134M per year and that cost has never been allocated anywhere. IT infrastructure, Finance operations, HR — all consumed by the three operating divisions in very different proportions, all sitting in a cost centre that reports zero revenue and zero allocated cost.
When you allocate that $134M using principled, driver-based methods — headcount for Finance and HR, square footage for Facilities, IT support tickets for IT — the picture changes completely.
The real decision impact: NovaPrism's CFO is considering a $200M capital investment to expand Infrastructure Solutions — its "highest-revenue" division. After proper allocation, IS margin is 3.0%. Cloud Licensing, at 40.3% margin, is the division that actually deserves capital. The blind spot doesn't just distort reporting — it misdirects investment.
Additionally, 17 of NovaPrism's 60 client engagements are loss-making on a fully-loaded basis. On direct costs alone, none appear to lose money. The pricing team has been renewing these contracts without knowing they destroy value every quarter.
Why spreadsheets fail here: The allocation logic is not complex — it is the governance that fails. Different analysts use different headcount figures. The square footage data lives in Facilities' own spreadsheet. IT ticket counts are pulled manually from the helpdesk system. Every quarter the numbers are slightly different. Nobody can audit the prior quarter's allocation because the file has been overwritten. EPCM solves the governance problem, not just the calculation problem.
The four costing verbs
The word "allocate" is used for everything. It shouldn't be. The verb you choose determines the methodology — and the degree of actionability the result provides.
Direct
The cost is directly traceable to a cost object from the source system. No calculation required — the data arrives assigned.
NovaPrism: Digital Services consultant salary costs. Each consultant's time is billed to a specific engagement. The cost lands on the right engagement automatically.
Highest actionability
Assign
A causal driver exists. Costs are assigned using a driver that has a clear cause-and-effect relationship with consumption.
NovaPrism: IT support costs assigned by IT ticket count. Infrastructure Solutions raises 2,890 tickets vs Digital Services' 1,240. IS consumes more IT — the driver reflects that.
Causal relationship
Attribute
An indirect cause-and-effect relationship exists. The recipient has some degree of control, but the link is less direct.
NovaPrism: Facilities costs attributed by square footage. IS occupies 98,000 sq ft vs DS's 42,000. The link is reasonable but IS cannot reduce the cost tomorrow by shrinking its footprint.
Indirect relationship
Allocate
No cause-and-effect relationship exists. Costs are spread using a simple metric where the recipient has no control over the cost.
Corporate tax provision allocated by revenue percentage. IS cannot do anything to reduce its share of the tax bill. The allocation is informational only — it describes a burden, not a cause.
No actionability
Design principle: Start with Direct. Where Direct is not possible, look for a causal driver and Assign. Where Assign is too granular, Attribute. Reserve pure Allocate for costs where the recipient genuinely has no control — corporate overheads, regulatory levies. Most shared services costs have a driver if you look for one. NovaPrism uses Assign for IT and Finance/HR. Nothing uses pure Allocate if a driver exists.
Where should allocations live?
The GL, Planning, and EPCM all support some form of allocation. Choosing the wrong tool creates a maintenance problem that grows with every business change.
| Requirement | General Ledger | Planning | EPCM |
|---|---|---|---|
| Time-sensitive statutory close | Best fit | ||
| Simple, single-step allocations | OK | OK | |
| Extend dimensionality beyond GL segments | Possible | Best fit | |
| What-if scenario analysis | Possible | Best fit | |
| Business analyst–controlled logic | Possible | Best fit | |
| 100+ unique allocation rules | Difficult | Best fit | |
| Waterfall / sequentially dependent rules | Best fit | ||
| Complex logic / reciprocal relationships | Best fit | ||
| Critical audit trail & traceability | Best fit |
NovaPrism's statutory close allocations stay in the GL. The management profitability model — four cost pools, five driver types, 60 engagement dimensions, four scenario models — lives in EPCM. These are not competing tools. They serve different purposes in the same finance architecture.
EPCM licensing: EPCM is included in Oracle's Enterprise EPM bundle — not a separate licence add-on. If NovaPrism already runs Planning or FCCS on Oracle EPM Cloud, EPCM is available at no additional licence cost. This changes the build-vs-buy calculation significantly.
What EPCM is, architecturally
EPCM is not a renamed version of the on-premise predecessor. It is a new product — built by combining three components that did not exist together before June 2022.
Layer 1 — Allocation Engine
PCM Modeling Capabilities
Waterfall allocation logic, rule sets, allocation and custom calculation rules, parallel / serial / iterative execution, model documentation, rule balancing. The core allocation engine — now rebuilt on ASO.
+
Layer 2 — Shared Infrastructure
EPM Platform
Metadata Management, Data Forms, Dashboards, Navigation Flows, Roles & Security, Audit Logs, Data Exchange, EPM Integration Agent, Data Maps, Connections, Oracle Analytics Cloud integration. The same platform shared by Planning, FCCS, and every Oracle EPM Cloud application.
+
Layer 3 — Architecture Enhancements
Dual-Cube Design
Two prebuilt ASO cubes replace the single-cube legacy architecture. Up to three additional ASO cubes can be added for driver staging, historical archiving, or reporting extensions. Released June 2022.
↓
Result
EPCM — Enterprise Profitability and Cost Management
A governed, auditable, business-user-managed platform for multi-dimensional profitability and cost allocation. Not a spreadsheet replacement — an enterprise allocation engine with a full platform beneath it.
The two prebuilt cubes
The most important architectural fact about EPCM is the separation of the calculation environment from the reporting environment. This is not just a performance choice — it is a governance choice.
PCM_CLC
Calculation Cube
Where allocations run. Active calculation environment. Business users doing what-if analysis work here.
- Allocation rules execute against this cube
- Working and Final versions live here
- Rule Balancing reads from here
- Not exposed to reporting users during active calculation
- Can be cleared and recalculated without touching PCM_REP
PCM_REP
Reporting Cube
Where results are published. The CFO, division heads, auditors — read from here.
- Populated via Copy POV from PCM_CLC after calculation
- Read-only for reporting users
- Historical POVs archived here
- Smart View connects to this cube for ad hoc
- Stable while the next period's allocation is being calculated
The legacy PCM architecture for comparison: The on-premise predecessor used a single ASO database. Calculation and reporting happened in the same cube. When the allocation recalculated, reporting users saw incomplete or intermediate results. EPCM's dual-cube design eliminates this. The two cubes are the architectural upgrade that makes EPCM production-grade for monthly close.
AI-first by design
EPCM is not a reporting tool that happens to have an AI button. The AI capability is meaningful only when the underlying model is correct — and building that model is what this tour is about.
🤖 What the AI layer can answer — once the model is built
"Which of NovaPrism's 60 client engagements lost money in Q1 FY24 on a fully-loaded basis?" — The AI queries PCM_REP, filters engagements where PCM_Output exceeds revenue, returns a ranked list with margin percentages. No form design needed. No Smart View layout needed. The question is answered in natural language against the calculated model.
The important caveat: the AI layer is only as good as the model beneath it. If NovaPrism's IT cost pool is allocated by revenue percentage instead of by ticket count, the answers will be confidently wrong. A poorly designed model produces wrong answers faster. This is why every hall in this tour is about model design first, AI consumption second.
The Academy principle: Every design decision in Halls 2–8 will be framed against the question "does this make the model's output trustworthy enough for an AI agent to reason against?" Trustworthy output requires correct driver design, proper PCM_Balance management, and a model that balances. Dean Eliot in the sidebar can be asked anything about NovaPrism's model at any time.
Day one at NovaPrism — the before state
You have just joined NovaPrism Group as the EPCM implementation lead. Your first task: document exactly what exists today and what is broken.
Mission Brief — Document the pre-EPCM allocation state
Situation: NovaPrism's Finance team has been running Q1 profitability using a shared Excel workbook maintained by one senior analyst. The workbook imports trial balances from four source systems, manually applies allocation percentages that were last formally reviewed 18 months ago, and produces a divisional P&L. Nobody else fully understands the logic. The analyst is leaving next month.
Identify the four cost pools. From NovaPrism's Enterprise Support cost centre — list the four pools (IT $52M, Facilities $38M, Finance $28M, HR $16M), their amounts, and the driver currently being used for each. Note whether each current driver reflects a cause-and-effect relationship or a simple spread.
Calculate the current misallocation. If Finance currently allocates all $134M by revenue percentage (DS 37.1%, IS 46.7%, CL 16.2%), compute the resulting allocated amount per division. Compare to the correct driver-based allocation. Which division is over-charged? Which is under-charged? By how much?
State the governance failures. List the specific governance problems the current Excel approach creates — auditability, version control, driver maintenance, what-if capability. For each, state what EPCM provides instead. This becomes the business case document for the implementation project.
Apply the four costing verbs. For each of NovaPrism's four cost pools, assign the correct costing verb: Direct, Assign, Attribute, or Allocate. Justify each choice. If a better driver than the current one exists, name it. This document is the design brief for Hall 4.
Write the CFO brief. One paragraph. What is the current reported margin for Infrastructure Solutions? What is the correctly-allocated margin? What capital allocation decision is currently being made incorrectly as a result? What is the risk if the decision is not corrected? This is the document that justifies the EPCM implementation to the executive sponsor.
0 of 5 tasks completed
Dean's hint: For task 2, the correct driver-based allocation produces DS: $40.26M, IS: $80.82M, CL: $12.86M. If Finance allocated by revenue percentage instead, IS would receive $62.6M — undercharged by $18.2M. DS would receive $49.7M — overcharged by $9.5M. Cloud Licensing would receive $21.7M — overcharged by $8.9M. These errors compound every quarter.
What Foundations Hall covers on the exam
Hall 1 maps to the first syllabus topic: Setting Up Profitability and Cost Management. Two sub-bullets. Expect 4–6 questions in this area across the 50-question exam.
📝 Syllabus — Setting Up Profitability and Cost Management
Sub-bullet 1: Describe Enterprise Profitability and Cost Management applications.
Know the three-component architecture (PCM + EPM Platform + Architecture Enhancements). Know the two prebuilt cubes and what each does. Know that EPCM Standard uses ASO — not BSO. Know that EPCM was released June 2022.
Sub-bullet 2: Create an application.
Know the application types available (Standard vs Custom). Know what is configured at creation and what cannot be changed later. Know who can create an application (Service Administrator only).
Know the three-component architecture (PCM + EPM Platform + Architecture Enhancements). Know the two prebuilt cubes and what each does. Know that EPCM Standard uses ASO — not BSO. Know that EPCM was released June 2022.
Sub-bullet 2: Create an application.
Know the application types available (Standard vs Custom). Know what is configured at creation and what cannot be changed later. Know who can create an application (Service Administrator only).
⚠️ Exam trap — the most common wrong answer in this section
"EPCM Standard uses BSO (Block Storage) for its calculation engine."
This is wrong. EPCM Standard uses ASO (Aggregate Storage). This matters because ASO and BSO have fundamentally different dimension design rules, calculation methods, and performance characteristics. Every dimension design decision in Hall 2 flows from the fact that it is ASO. The exam will test this. Do not confuse EPCM with Planning (which uses BSO for its calculation cube).
This is wrong. EPCM Standard uses ASO (Aggregate Storage). This matters because ASO and BSO have fundamentally different dimension design rules, calculation methods, and performance characteristics. Every dimension design decision in Hall 2 flows from the fact that it is ASO. The exam will test this. Do not confuse EPCM with Planning (which uses BSO for its calculation cube).
⚠️ Exam trap — the most common wrong answer in this section
"EPCM Standard uses BSO (Block Storage) for its calculation engine."
This is wrong. EPCM Standard uses ASO (Aggregate Storage). This matters because ASO and BSO have fundamentally different dimension design rules, calculation methods, and performance characteristics. Every dimension design decision in Hall 2 flows from the fact that it is ASO. The exam will test this. Do not confuse EPCM with Planning (which uses BSO for its calculation cube).
This is wrong. EPCM Standard uses ASO (Aggregate Storage). This matters because ASO and BSO have fundamentally different dimension design rules, calculation methods, and performance characteristics. Every dimension design decision in Hall 2 flows from the fact that it is ASO. The exam will test this. Do not confuse EPCM with Planning (which uses BSO for its calculation cube).
🔐 Security model — two layers, one exam topic
The 1Z0-1082-25 exam asks multi-select security questions: "You want to limit the slices of data users can view. Which two mechanisms can you use?" The answer is always a combination from the two-layer model below.
Layer 1 — Predefined roles (identity domain level)
Four roles, hierarchical — each inherits everything below it:
Layer 2 — Data grants (application level)
Data grants restrict which data slices a User or Viewer can see — e.g. only their own division's costs, only their region, only specific engagements. Created by Service Administrators and assigned to native groups (not to predefined groups directly).
Key rules the exam tests:
• Service Administrators and Power Users always see all data — data grants cannot restrict them
• Data grants apply only to User and Viewer roles
• Assign data grants to native groups, not directly to predefined groups
• Privileges are additive only — application roles can extend but never reduce a predefined role's access
• Application roles (e.g. Data Integration - Create) can grant Users extra capabilities beyond their predefined role
The exam answer to "which two mechanisms limit data access?"
→ Application roles (extend functional access) + Security filters / Data grants (restrict data slices). User groups alone do not restrict data — they are a container for assigning grants efficiently.
Layer 1 — Predefined roles (identity domain level)
Four roles, hierarchical — each inherits everything below it:
| Role | What they can do in EPCM | NovaPrism |
|---|---|---|
| Service Administrator | Full access. Create/administer application, manage security, create data grants, assign roles. Only role that can create an EPCM application. | Implementation lead |
| Power User | Create POVs, load data, create/run rules and calculations, define profit curves, run Rule Balancing. Access to all data — cannot be restricted by data grants. | Finance Controller |
| User | Analyse data, create reports and dashboards, view profit curves. Can be restricted to specific data slices via data grants. | Division analysts |
| Viewer | View reports and dashboards only — read-only. Can be restricted via data grants. | CFO, board recipients |
Data grants restrict which data slices a User or Viewer can see — e.g. only their own division's costs, only their region, only specific engagements. Created by Service Administrators and assigned to native groups (not to predefined groups directly).
Key rules the exam tests:
• Service Administrators and Power Users always see all data — data grants cannot restrict them
• Data grants apply only to User and Viewer roles
• Assign data grants to native groups, not directly to predefined groups
• Privileges are additive only — application roles can extend but never reduce a predefined role's access
• Application roles (e.g. Data Integration - Create) can grant Users extra capabilities beyond their predefined role
The exam answer to "which two mechanisms limit data access?"
→ Application roles (extend functional access) + Security filters / Data grants (restrict data slices). User groups alone do not restrict data — they are a container for assigning grants efficiently.
0/5
Hall 1 Exam Score
| Concept | Exam answer | Implementation reality |
|---|---|---|
| Storage engine | ASO (Aggregate Storage) | Architecturally significant — drives all dimension design decisions in Hall 2 |
| Prebuilt cubes | PCM_CLC (calculation) + PCM_REP (reporting) | You reference these cube names in every data integration and export task |
| Additional cubes | Up to 3 additional ASO cubes | Most implementations use 1 — for driver staging. 3 is the ceiling, not the default |
| EPCM release | June 2022 | Relevant for migration questions — anything built before June 2022 is legacy PCM |
| Who creates an application | Service Administrator | In practice, the implementation consultant creates in Sandbox first, then promotes to PROD |