Analyse small suppliers

# Tail spend analysis (long tail)

## Purpose
Apply the Pareto principle to supplier spend: identify "tail spend" (the many small suppliers representing little volume but generating a lot of operational overhead), and find consolidation opportunities.

## Instructions

### Step 1 — Retrieve spend by supplier over 12 months
```
Call spendesk_analyze_spend with:
  - groupBy: ["supplier"]
  - limit: 50 (by default only 10 rows come back; the rest is rolled into an "Others" row)
  - Period: last 12 months (366 days at most per call)
  Each row gives the amount and the number of expenses (count), plus an "Others" row.
  ⚠️ The long tail sits precisely in "Others": its amount and count are known, but not the
  supplier-by-supplier detail. Say so, and only detail the listed suppliers.
  Compute shares on summary.totalAmount, not on the share field.

Result: list of suppliers with annual volume.
Sort by descending amount.
```

### Step 2 — Calculate the Pareto curve
```
Total spend = sum of all amounts

For each supplier (sorted by amount desc):
  Calculate % of total and cumulative %

Identify the thresholds:
  Tier A (strategic): suppliers representing the first 70% of cumulative volume
  Tier B (preferred): suppliers from 70% to 90% of cumulative volume
  Tier C (tail spend): suppliers representing the remaining 10% of volume

Typically: 80% of volume is concentrated on ~20% of suppliers.
```

### Step 3 — Analyze tail spend
```
For Tier C suppliers:
  Count: X suppliers | X€ total | X% of spend
  Average amount per supplier
  Number of transactions per supplier

Tail spend sub-categories:
  💲 One-shot: a single purchase, never renewed
  🔄 Micro-recurring: < 5 transactions/year, < 500€/year
  📦 Consolidatable category: same type of service as Tier A/B
```

### Step 4 — Identify consolidation opportunities

⚠️ **Response-size risk — critical here**: Tier C is by definition the long tail, potentially hundreds of suppliers. `get_suppliers` embeds an `accountPayable.supplierIds` field that can blow up response size for any supplier sharing a consolidated AP account — even `pageSize=3` can fail. **Never call `get_suppliers` across the whole Tier C.** Only target the 5-10 Tier C suppliers with the most transactions, via `ids=[...]`, one at a time or in very small batches.

⚠️ `get_suppliers` has no "industry" field — only `name`, `legalName`, `address`, banking details. Matching "similar suppliers" stays manual, based on payable descriptions (`get_payables[].description`) rather than a structured field.

```
For tail spend, look for duplicates by category:
  Ex: 12 training suppliers → can we consolidate to 1-2 partners?
  Ex: 8 travel agencies → consolidate to a single one?

Call get_suppliers with ids=[...] for the 5-10 Tier C suppliers with the most transactions (targeted, never broad pagination):
  Check: name, country
  Cross-reference payable descriptions (get_payables) to guess the service type
```

### Step 5 — Tail spend report
```
Present:
1. Pareto table:
   Tier | Nb suppliers | Total spend | % of total
   A (≥ 70%) | X | X€ | 70%
   B (70-90%) | X | X€ | 20%
   C (tail, <10%) | X | X€ | 10%

2. Consolidation opportunities:
   Category | Current nb suppliers | Total spend | Suggested action

3. One-shot suppliers > 6 months (no recent activity):
   To archive to clean up the master data

4. Estimated potential savings from consolidation:
   Reduced admin overhead + volume negotiation
```

## Points of attention
- Tail spend is not necessarily bad — some one-off purchases are unavoidable
- Consolidation must be validated by the business (changing suppliers impacts operations)
- Some "small" suppliers may be strategic (niche, rare quality)

## Example prompts
- "How many suppliers represent 80% of my spend?"
- "Are there consolidation opportunities in my tail spend?"
- "Show me the suppliers I only use once a year"