Insurance Renewal LINE Bot PM Case Study

B2B tool PM practice | PRD assumption validation and requirement recalibration

A renewal-management system for a small Taiwanese property-insurance brokerage office. The focus is not only the final PRD, but the key recalibration step and the reasoning behind it.

Timeline
Apr 2026
Delivery
3 personas · 10 functions · 3 ADRs
Customer
Small Taiwanese insurance brokerage office
Market Size
NT$10-33M / year
NestJSSupabaseLINE Bot SDKPostgreSQLThe Mom Test
Screenshots of the Insurance Renewal LINE Bot for Admin, Agent, and Staff roles with sensitive data blurred
LINE Bot running in practice

Admin global entry, Agent claim flow, and Staff sheet-sync preview with sensitive information blurred. This is evidence that the responsibility-flow design can actually run.

Case Brief
This was not simply a LINE Bot. It was a vague renewal-reminder need recalibrated into a responsibility-flow system.

My focus was user interviews, product positioning, MVP scope, pricing estimation, and ADR decisions.

PM Evidence
  • Used The Mom Test to challenge PRD v1.0 assumptions
  • Split scope into Phase 1 MVP and explicit non-goals
  • Built personas, data model, technical stack, and pilot timeline
  • Used ADRs to preserve key decision context, tradeoffs, and consequences
433
Insurance brokerage firms
Public FSC data
5-15
Typical small-office headcount
3-8K
Estimated monthly ACV
B2B SaaS range
10-33M
Potential market NT$ / year
Not unicorn-scale, but real

The problem that triggered this project

We often just skim the email and register the policy, then something gets missed. We had a policyholder whose agent did not proactively follow up, and the customer also missed the insurer's notice. The property-insurance policy lapsed, and it became a complaint.

From an interview with a small brokerage-office owner

Insurance renewals theoretically have two notification lines: the insurer notifies the policyholder before expiration, and the broker agent should proactively contact the customer. In practice, both lines can fail: the agent does not follow up, and the policyholder ignores the insurer's SMS, email, or paper notice. The policy truly lapses. At best it becomes a complaint; at worst, a claim during the lapse period cannot be paid.

This is not an occasional mistake. It puts responsibility on two unreliable human links: "the agent remembers" and "the policyholder notices." I judged this to be a common small-office pattern worth solving with a system.

Why this small market was worth doing

Fragmented market
01 · No one owns this niche
Hundreds of small offices operate independently. The market is too small and fragmented for large SaaS players.
Legal notice is not enough
02 · Insurers notify but notices get ignored
SMS, email, and paper notices can all be missed. Broker follow-up is the critical defense line.
AI landing zone
03 · Sequoia AI thesis fit
Insurance brokerage has high knowledge intensity and fragmented workflows, making it a strong AI application market.

Competitive landscape and the gap I saw

Competitor / positionCharacteristicsLimitation for small offices
PIRANS renewal helperLINE Bot plus Google Sheets, positioned as an individual-agent tool.Validated willingness to pay, but has no team collaboration mechanism.
Franchise brokerage back officesSystems from major brokerage networks are feature-rich.Designed for headquarters managing branches, not for collaboration inside one small office.
Gap I sawTurn 'contact window → staff → agent → customer' into enforced responsibility flow.Fill the space between PIRANS and franchise systems.
Core project insight

This case is not about the final PRD alone. It is about the key self-correction in the middle: how I treated PRD v1.0 as a hypothesis, then used The Mom Test-style interviews to reposition the product from a renewal reminder tool into a responsibility-flow system.