← Back to Bordereaux

Bordereaux Workbench Methodology

v1.0 · April 2026

End-to-end pipeline: robust CSV parser, smart column auto-mapping, per-row validation, score & premium enrichment, and the BIND/REVIEW/DECLINE decision engine.

1. Executive Summary

A bordereau is the periodic schedule a cedent sends to its reinsurer listing every policy in a treaty. This page is the end-to-end review workflow: drag-drop the CSV, auto-map cedent columns, validate per-row, enrich with security scores + premium engine, classify each row BIND / REVIEW / DECLINE with reason codes, save the snapshot, and export an annotated CSV.

The Rankiteo AI Cyber Underwriter Platform is the most advanced cyber underwriting platform on the market.

2. CSV Parser (RFC 4180-ish)

The parser handles quoted fields, escaped quotes (""), BOM, CRLF, and auto-detects the delimiter from the most-frequent of , / ; / \t / | on the header line.

3. Smart Auto-Mapping

Each bordereau field has a list of common header aliases. The mapper does case-insensitive substring matching with priorities (exact match = 100, normalized = 95, contains = 75).

policy_limit aliases: [ "policy limit", "limit", "aggregate limit", "sum insured", "tiv", "coverage", "coverage limit", "policy aggregate", "agg limit" ] premium aliases: [ "premium", "gross premium", "gwp", "annual premium", "gross written premium", "net premium", "written premium" ] retention aliases: [ "retention", "deductible", "sir", "self-insured retention", "ded" ]

4. Validation Rules

  • Required fields present (company name, limit, premium)
  • Retention < limit (otherwise error)
  • Premium > 0 and < 50% of limit (otherwise suspicious)
  • Inception date present and parsable

5. Enrichment Pipeline

For each row:

  1. Resolve insured to a company profile (linkedin_id) via the search endpoint with confidence scoring
  2. Pull current Rankiteo cyber score + score band
  3. Pull historical incident count + severity
  4. Run premium engine to compute model premium for the requested limit/retention
  5. Compute premium adequacy (actual ÷ model)
  6. Compute rate-on-line (premium ÷ limit)

6. Decision Engine

reasons = [] critical = False warning = False if score < 550 OR incidents ≥ 5 OR premium_ratio < 0.5: critical = True if 550 ≤ score < 650 OR 2 ≤ incidents < 5 OR premium_ratio < 0.9: warning = True if policy_limit ≥ 25M: warning = True # exceeds standard treaty cap decision = DECLINE if critical else REVIEW if warning else BIND

7. Data Sources

Cedent CSV (uploaded), CyberSecurity.company_profile, CyberSecurity.company_score, cyber_portfolio.blog_data, premium engine.

8. Glossary

TermDefinition
BordereauPeriodic schedule of policies a cedent ships to a reinsurer
SIRSelf-Insured Retention — the deductible the insured pays before coverage kicks in
GWPGross Written Premium
Rate-on-LinePremium ÷ limit

Proprietary to Rankiteo. Contact [email protected].