PRD Template for Insurance Products

Example: "Auto insurance claims take an average of 18 days from first notice of loss (FNOL) to settlement payment. 34% of this time is spent on manual document collection: police reports, repair estimates, medical records, and photos. Policyholders submit an average of 4.2 follow-up inquiries during the claims process, each costing $12 in call center time. Competitors offering AI-assisted claims processing report 7-day average resolution with 22% higher customer satisfaction scores."

Example: "Primary: Reduce average FNOL-to-settlement time from 18 days to 9 days within 6 months. Secondary: Reduce claims handling cost per claim from $420 to $280. Policyholder satisfaction: Increase claims NPS from 28 to 45. Compliance: Maintain 100% compliance with state-mandated claims acknowledgment and response timelines. Accuracy: Maintain or improve claims accuracy — leakage rate must not increase above current 4.2% baseline."

Example: "As a policyholder who has just been in a car accident, I want to file a claim from my phone at the scene so that the claims process starts immediately without waiting for business hours. Acceptance criteria: claim filed in under 5 minutes via mobile app; supports photo upload of damage, police report number, and other party information; real-time confirmation with claim number and assigned adjuster name; automated acknowledgment email sent within 1 hour per state requirements."

Example: "FR-1: FNOL intake must capture: date/time of loss, location, description, involved parties, injuries, police report number, and up to 20 photos. FR-2: AI triage must classify claims into fast-track (estimated settlement under $5,000) and standard within 30 seconds of submission. FR-3: Fast-track claims must auto-generate repair estimate using photo AI analysis with accuracy within 15% of adjuster estimate. FR-4: Claims payment must support ACH direct deposit, check, and repair shop direct pay. Payment initiated within 24 hours of claim approval. FR-5: All claims decisions must include documented rationale auditable by state regulators."

Example: "CR-1: Claims acknowledgment within 15 days of FNOL in all states (some states require acknowledgment within 24 hours — build to the strictest standard). CR-2: Claims decision communicated to policyholder within 30 days (varies by state; California requires 40 days). CR-3: All claims denials must include specific policy language justifying the denial and information about the appeals process per state requirements. CR-4: Rate filing compliance: any algorithmic underwriting factors must be filed with and approved by state insurance departments before use. CR-5: NAIC Market Conduct compliance — maintain records demonstrating fair claims handling practices."

Example: "Speed: Average FNOL-to-settlement time. Baseline: 18 days. Target: 9 days. Cost: Claims handling cost per claim. Baseline: $420. Target: $280. Satisfaction: Claims NPS. Baseline: 28. Target: 45. Accuracy: Claims leakage rate (overpayment as % of total claims paid). Baseline: 4.2%. Target: under 3.5%. Fraud: AI-flagged suspicious claims as percentage of total. Target: identify 90% of confirmed fraud cases. Compliance: Percentage of claims meeting state-mandated response timelines. Target: 100%."

Example: "Phase 1 (Months 1-2): FNOL mobile intake with photo upload and AI triage. Milestone: Triage accuracy validated against 1,000 historical claims. Phase 2 (Months 3-4): Fast-track auto-settlement for claims under $5,000. Milestone: Legal and compliance review of auto-settlement authority and disclosure language. Phase 3 (Month 5): Pilot with one state and one line of business. Milestone: 500 claims processed with quality audit. Phase 4 (Months 6-8): Expand to additional states, incorporating state-specific compliance rules. Phase 5 (Month 9): Full rollout across personal auto book."

Example: "Risk: AI triage misclassifies a complex bodily injury claim as fast-track, leading to inadequate investigation and underpayment. Likelihood: Low. Impact: Critical (regulatory action + bad faith lawsuit). Mitigation: Bodily injury claims always routed to human adjuster regardless of AI classification; AI confidence threshold of 95% required for fast-track; weekly audit of AI triage decisions by claims supervisor. Risk: State insurance department objects to AI-assisted claims decisions during market conduct exam. Likelihood: Medium. Impact: High. Mitigation: Maintain full decision audit trail; ensure all AI recommendations are reviewed by licensed adjuster; document AI model methodology for regulatory filing."

Example: "Integration 1: Policy admin system (Guidewire PolicyCenter/Duck Creek) — real-time policy and coverage verification. Integration 2: Claims management system (Guidewire ClaimCenter) — claims lifecycle management and payment processing. Integration 3: External data — LexisNexis (claims history), weather data (CAT event correlation), telematics (driving behavior for auto). Integration 4: Payment processing — ACH disbursement via carrier treasury system. Integration 5: Document management — integration with carrier document repository for claims file maintenance."

Example: "Q1: What is the maximum claim amount that can be auto-settled without human adjuster review? Legal and actuarial input needed. Decision owner: Chief Claims Officer. Needed by: Month 2. Q2: Should we build our own AI claims triage model or license from a vendor (Tractable, CCC)? Build gives us differentiation but takes longer. Decision owner: CTO. Needed by: Month 1. Q3: How do we handle claims in states where AI-assisted claims processing requires specific regulatory disclosure? Legal review of 50-state landscape needed. Decision owner: Chief Compliance Officer. Needed by: Month 3."