Spend per employee by department

# Spend per employee — per capita analysis

## Purpose
Calculate and analyze average spend per employee by cost center. Identify teams whose spend per head is significantly above or below the internal average, and track how this KPI evolves over time.

## Instructions

### Step 1 — Retrieve active users by cost center
```
Call get_users (with pagination).
For each active user: note id, costCenterId, role, status.
Count the number of active users per costCenterId → headcount.

Note: get_users returns Spendesk users (not necessarily the entire company).
Headcount is therefore the number of Spendesk users per team.
```

### Step 2 — Retrieve spend by cost center
```
Call spendesk_analyze_spend with:
  - groupBy: costCenter
  - Period: current month (or M-1 for complete data)
  - Include all expense categories

For each cost center: id, name, totalAmount, currency.
```

### Step 3 — Calculate spend per capita
```
For each cost center with headcount > 0:
  spendPerEmployee = totalAmount / headcount

Global average = totalSpend / totalHeadcount

For each center:
  Ratio vs average = spendPerEmployee / global_average

Categories:
  🔴 > 2x the average: very high spend per capita
  🟡 1.5x–2x: above average
  🟢 0.5x–1.5x: normal range
  💡 < 0.5x: underutilization (team not very active on Spendesk?)
```

### Step 4 — Breakdown by category (top 5 centers)

⚠️ **Verified bug**: `expenseCategory` does not exist as a `spendesk_analyze_spend` dimension — calling it this way returns a server error (`"grouper is not a function"`). Valid dimensions are: `costCenter, supplier, employee, month, week, currency, analyticalCode, payableType, bookkeepingStatus, paymentStatus, vatRate`. In this Spendesk instance, employee-facing "expense categories" are carried by the custom analytical field "Analytic codes" — use the `analyticalCode` dimension instead:

```
For the 5 cost centers with the highest spend per capita:
  Call spendesk_analyze_spend with groupBy: [costCenter, analyticalCode], filters.costCenterIds=[<center id>]
  analyticalCode values already come back named in the row's breakdown[] — no need for get_analytical_field_values unless you need to double-check a raw ID
  Identify the top 3 categories explaining the spend level
```

### Step 5 — Evolution over 3 months
```
Calculate spend per capita for the last 3 months for the 5 largest centers:
  Compare: trend rising / stable / declining?
```

### Step 6 — Per capita report
```
Present:
1. Global KPI: X€ per employee / month (Spendesk)
2. Table by cost center:
   Team | Headcount | Total spend | Spend/employee | vs Average
3. Top 3 teams with high spend per capita (with category breakdown)
4. Fast-growing teams (3 months)
5. Recommendations: which teams deserve a spend review?
```

## Points of attention
- Spendesk headcount ≠ total HR headcount — some employees don't have a Spendesk account
- Teams with specific missions (field sales, events) naturally have higher levels
- Interpret in business context before drawing conclusions

## Example prompts
- "Which team has the highest spend per person?"
- "Calculate average spend per employee by department"
- "How has our per capita spend evolved over the last 3 months?"