---
title: CFO Board
description: "LLM prompt execution cost allocation, token telemetry, and custom tenant model rate management."
---

The **CFO Board** provides financial oversight and cost control for Large Language Model (LLM) usage across your organization. It allows financial officers and administrators to track prompt execution volumes, analyze token usage (input, completion, cached, and reasoning tokens), evaluate costs per team and model, and configure custom tenant model rates.

## Plan & Access Requirements

Access to the CFO Board is restricted based on tenant subscription and user capability:

- **Plan Requirement:** **Premium** or **Enterprise** tenant plan (not available on Standard plan).
- **User Role:** Must be a **Superadmin**.
- **User Capability:** Must possess the `finance` capability assigned by tenant administrators.

---

## CFO Telemetry Pipeline

The CFO Board relies on the **CFO Telemetry Pipeline** to aggregate prompt execution metrics without storing sensitive prompt contents or Personally Identifiable Information (PII).

### Privacy & Telemetry Design

- **Execution Header (`X-PE-Execution-ID`)**: When a prompt is fetched via the REST API or MCP tools (such as `get_prompt`), the system generates a unique execution ID returned in the response header.
- **Zero Content Storage**: Only numeric token counts (prompt, completion, cached, reasoning tokens), LLM model patterns, execution status, latency, and team ID are stored in the telemetry summaries.
- **Opt-In Switch**: Telemetry collection can be paused or activated at any time from **CFO Board -> Settings**. When paused, no telemetry requests are logged, and statistics display zero placeholder values.

---

## Key Metrics & Statistics

The **Statistics & Telemetry** tab provides key performance indicators (KPIs) and interactive visualizations:

### Summary Cards
- **Total Estimated Cost**: Total estimated expenditure calculated by matching prompt execution token counts against your tenant model cost rates.
- **Prompt Executions**: Total count of prompt calls with completed and failed execution breakdowns.
- **Input & Cached Tokens**: Total prompt tokens sent to LLM providers, including prompt caching metrics.
- **Output & Reasoning Tokens**: Total completion tokens received from LLM providers, including specialized reasoning/thinking tokens.

### Daily Trends Chart
An interactive Recharts visualization graphing daily cost accumulation (€) or execution volume over a selected date range.

### Breakdown Tables
- **Cost Allocation by Team**: Displays prompt execution volume, percentage share, and total estimated cost grouped by internal teams.
- **Usage & Cost by LLM Model**: Displays execution volume, input/output token counts, cached tokens, and estimated cost grouped by LLM model identifiers (e.g. `claude-3-5-sonnet-20241022`, `gpt-4o`, `gemini-1.5-pro`).

---

## Tenant Model Cost Rates

From **CFO Board -> Settings -> Tenant Model Cost Rates**, administrators can define custom LLM token pricing for exact cost allocation.

### Rate Configuration Parameters
- **Model Pattern**: Regex or exact model pattern (e.g., `claude-3-5-sonnet`, `gpt-4o`, `gemini-1.5-*`). Use `*` as a default fallback pattern.
- **Token Unit Size**: Base unit size for pricing calculation (default: `1,000,000` tokens / 1M).
- **Prompt Token Cost**: Cost in cents per token unit for input tokens.
- **Completion Token Cost**: Cost in cents per token unit for output tokens.
- **Cached Prompt Token Cost**: Reduced cost in cents per token unit for cached input tokens.
- **Reasoning Token Cost**: Cost in cents per token unit for reasoning/thinking tokens.

### Pattern Priority & Unique Constraint
- **Longest Match Priority**: When calculating execution costs, the system evaluates configured rates and selects the longest matching model pattern.
- **Unique Constraint**: Each tenant can configure only one rate pattern for a specific model string, preventing ambiguous cost evaluations.

---

## Audit Logging

All administrative actions on the CFO Board are recorded in the tenant [Audit Log](https://prompteasy.eu/docs/administration/audit-log.md):

- `CFO Board: toggle` (Activating or pausing the CFO Telemetry Pipeline)
- `Tenant Model Rates: add` (Adding a new model pricing pattern)
- `Tenant Model Rates: edit` (Updating an existing model pricing pattern)
- `Tenant Model Rates: remove` (Deleting a model pricing pattern)