Agent Portal for an Insurance Company
Hundreds of agents working blind: calculating premiums from outdated tables, no client history, waiting days for an underwriter. We explain how to build an agent portal with automated underwriting — policy issuance in 15 minutes, loss ratio down 18%.
A typical situation in insurance companies: hundreds of agents across the country, growing sales — but with them, growing loss ratio. Agents work autonomously: calculating premiums from Excel tables updated quarterly, no access to client history, coordinating non-standard cases with the underwriter by phone. Premium calculation errors reach 10–15%. Management blames agents, but the problem is the tools.
Why does this happen: agents work blind. They have no access to current tariffs, client claims history, or object scoring. The core system holds all this information but is locked behind the office perimeter. The underwriter — the only person with the complete picture — reviews every policy equally: both a standard auto policy for $300 and a complex industrial risk for $50K. The result: the underwriter is overloaded, standard policies wait 2–3 days, agents in the field lose clients.
Our approach to this challenge: the most labor-intensive and most valuable stage is formalizing underwriting rules. We sit with senior underwriters and work through 150–200 real cases: why this policy was approved, why that one was declined, what factors influence the decision. It turns out that 70–75% of decisions follow a few simple rules that underwriters apply intuitively. We translate these rules into an algorithm — standard policies are issued automatically in minutes. The underwriter focuses on the remaining 25–30% — complex and large risks where their expertise is truly needed.
The practical outcome: portfolio loss ratio drops 18% — not because the algorithm is smarter than a human, but because it applies rules consistently to every policy, without fatigue and without exceptions. Agents get a mobile app with current tariffs, client history, and instant calculation. Premium errors drop from 12% to 1.5%. And an unexpected bonus: when an agent issues a policy in 15 minutes instead of two days, they can serve three times as many clients — sales grow too.
Typical Problem
A typical picture in insurance companies with large agent networks: agents calculate premiums from Excel tables updated quarterly. No access to client history. Non-standard cases are coordinated with the underwriter by phone — wait 2–3 days. Every policy passes through 4–5 hands, premium calculation errors reach 10–15%, portfolio loss ratio grows due to the absence of systematic risk assessment.
Why This Happens
The company grows through its agent network but manages it with 10–15-year-old tools. Three systemic problems: (1) agents are cut off from current tariffs and client history — working blind, (2) the underwriter reviews all policies equally — both a standard auto policy for $300 and a complex industrial risk for $50K, creating a bottleneck, (3) the core system accumulates data but doesn't use it for risk assessment. Management blames agents, but the problem is the absence of tools.
How We Diagnose It
The problem isn't the agents — they work blind, without access to current tariffs, claims history, or scoring data. The core system holds all the information but is locked behind the office perimeter. The underwriter — the only person with the complete picture — is overloaded with routine, reviewing every policy equally. Our approach: we formalize underwriting rules by working through real cases with senior specialists. It turns out that 70–75% of decisions follow a few simple rules. These can be automated, and the underwriter can focus on complex cases.
The Right Model
Agent portal with three levels: (1) premium calculator with current tariffs and core system connection for history checks, (2) automatic underwriting for standard risks based on rules, (3) routing of non-standard risks to the underwriter with pre-calculated scoring.
How We Implement It
We design the agent portal, integrate with the core system via API. We formalize underwriting rules — working through 150–200 real cases with senior underwriters to identify actual decision-making logic. We develop a scoring model based on claims history. We launch a mobile app for agents, configure a loss ratio dashboard by segment. Formalizing underwriting rules is the most labor-intensive stage, but also the most valuable. A typical project takes 5–7 months.
How the Team Works
Projects like this run with a team of 6: 2 developers, 1 business analyst, 1 actuary (from the client side), 1 integration engineer, 1 tester. I define the portal architecture, auto-underwriting model, and core system integration. The team implements, tests, and documents.
Results
If your underwriters make decisions 'by experience' and review every policy equally — loss ratio will grow. Formalize the rules, automate standard cases, and the underwriter will focus on where their expertise is truly needed.
Key Lessons
- • If your agents calculate premiums from Excel tables — errors are inevitable. Give them a tool with current tariffs and client history.
- • Automated underwriting reduces loss ratio not because it's smarter than a human, but because it applies rules consistently to every policy — without fatigue or exceptions.
- • Mobile app for agents is not a convenience but a necessity: the agent works 'in the field,' not at a desk. Without mobile access, they lose clients.
- • A loss ratio dashboard by segment gives management a tool they normally don't have — for the first time, they can see which products earn and which lose.
Similar challenge?
Tell me what's going on. I'll review the situation and suggest a concrete path forward.
Обычно отвечаю в течение нескольких часов