FRAI case study
Insurance Sales Workflow Case Study: 68% More Policy Sales
How HI Expats increased completed policy sales by 68% and cut qualification time per qualified lead by 38% with a managed sales workflow.
- Published
- Updated
- Reading time
- 7 min read
- Author
- Francesco Fuso
- +67.6% completed policy sales: 34 before and 57 after implementation
- +63.6% qualified leads: 176 before and 288 after implementation
- -37.5% average qualification time per qualified lead: 16 minutes to 10 minutes
- Qualification workload remained almost flat while the team processed 64% more qualified leads
- This is an observational before-and-after comparison, not a controlled experiment or proof of exclusive causation
- Live demo available — test the WhatsApp insurance chatbot now
How we measured the result
The figures below are internal operational KPIs from the HI Expats CRM dashboard. They compare two consecutive 92-day periods using the same metric definitions. The comparison was reviewed by FRAI on 16 July 2026.
| Measurement field | Baseline | After implementation |
|---|---|---|
| Dates | 1 March–31 May 2025 | 1 June–31 August 2025 |
| Incoming leads | 420 | 545 |
| Qualified leads | 176 | 288 |
| Completed policy sales | 34 | 57 |
| Average manual qualification time | 16 minutes | 10 minutes |
| Total qualification workload | 46.9 hours | 48.0 hours |
Metric definitions
A completed policy sale is a paid and issued insurance policy. Policies cancelled within 14 days, fully refunded policies, duplicate records, internal tests, incomplete applications, failed payments, and policies that were never activated are excluded.
A qualified lead is an enquiry that reached the agreed qualification status in the CRM after the information required for a meaningful review had been collected. Qualification time covers the manual time used to review, complete, and route that qualified enquiry before an adviser begins the sales conversation.
Calculations
1Completed policy sales increase2((57 - 34) ÷ 34) × 100 = 67.6%34Qualified leads increase5((288 - 176) ÷ 176) × 100 = 63.6%67Qualification-time reduction8((16 - 10) ÷ 16) × 100 = 37.5%910Baseline qualification workload11176 × 16 minutes = 2,816 minutes = 46.9 hours1213Post-implementation qualification workload14288 × 10 minutes = 2,880 minutes = 48.0 hours1516Work avoided at the previous handling rate17(288 × 16 minutes) - (288 × 10 minutes) = 28.8 hours
What changed, and what the comparison can show
Incoming lead volume increased from 420 to 545, or 29.8%, during the post-implementation period. Lead-to-sale conversion increased from 8.1% to 10.5%, while qualified-lead-to-sale conversion moved from 19.3% to 19.8%. The clearest operational result is therefore capacity: HI Expats processed 64% more qualified leads while measured qualification workload remained almost flat.
The FRAI workflow changed how enquiries were captured, completed, qualified, routed, and followed up. The comparison is observational, not a controlled experiment, and does not claim that software was the only cause of the sales increase. Demand, campaign mix, seasonality, pricing, staffing, and other commercial conditions can also affect sales.
Source, exclusions, and review status
• Data source: internal HI Expats CRM dashboard KPIs. • Measurement dates: 1 March–31 May 2025 and 1 June–31 August 2025. • Exclusions: cancelled, refunded, duplicate, test, incomplete, failed-payment, and never-activated policy records as defined above. • Client status: HI Expats is named in this case study. • Limitations: operational before-and-after comparison; no claim of exclusive causation. • Last reviewed: 16 July 2026.
Compliance. The workflow records opt-in, limits data collection to information needed for the quote, applies retention controls, and keeps an auditable interaction log.
Results at a glance & why it works
Available outside office hours
Prospects can start a WhatsApp conversation at any time. In the measured workflow, about 15% of conversations began after 20:30, when the office was normally closed.Fast first responses
The system answers common questions from a reviewed HI Expats knowledge base and uses the WhatsApp Cloud API to deliver the response. This gives prospects an immediate first contact while the human team remains responsible for advice and policy decisions.Knowledge-grounded answers
A large language model drafts answers using information retrieved from a reviewed knowledge base. This limits the source material available to the model, but it does not remove the need for monitoring or human review.What changed in practice (HI Expats case)
We implemented this flow for HI Expats, a specialist health‑insurance broker for international residents. Those are the steps decided and that are converting.
Answer questions, then qualify
The conversation begins by answering common questions. When a prospect shows clear buying intent—for example, asking about cover for a family moving to Spain—the workflow asks a small number of relevant questions to collect the context an agent needs.
Progressive profiling, not interrogation
Each answer adds context without forcing the prospect through a long form. When the prospect requests a quote, the workflow sends the appropriate form. After submission, the agent receives the structured form data and the relevant conversation context together.
Deterministic handoff
Freeing human agents from repetitive screening lets them focus on advice and sales conversations. Average manual qualification time fell from 16 to 10 minutes per qualified lead, a 37.5% reduction according to the internal CRM dashboard KPIs described in the measurement section above.
One channel for the first conversation
Prospects can read, reply, and provide the initial information in WhatsApp. The workflow then transfers the collected context to the team instead of asking the prospect to repeat it.
Auditable steps, human responsibility
The workflow records the automated steps and follows the defined routing rules. That traceability supports review and process improvement; it does not by itself guarantee legal or regulatory compliance. HI Expats remains responsible for advice, underwriting, and final insurance decisions.
Try the HI EXPATS WhatsApp insurance bot
Instant answers, progressive profiling, and a clean handoff to an agent.
- 01 · Free live demo · no signup
- 02 · Works on mobile and desktop
- 03 · Secure and GDPR-ready
Opens WhatsApp with a pre-filled message.
The Insurance Lead Qualification Process That Converts
A converting qualification flow does three things quickly: removes uncertainty, captures intent, and routes complete context to a licensed agent. Here’s the flow we’ve seen work reliably in insurance:
1) Answer‑first Q&A
Start with natural Q&A to remove friction. The bot uses your curated knowledge base to answer policy scope, eligibility, and documentation questions. This builds trust and reveals purchase intent.
2) Intent detection and light questions
When intent is explicit (e.g., “family of 3 moving to Spain—need health insurance”), switch to a few targeted questions (location, dependents, plan type, start date). Keep it conversational; avoid long forms.
3) Progressive profiling and consent
Collect only what’s needed for quoting (name, contact, basic risk attributes) and record opt‑in/consent for follow‑up. Respect policy and privacy constraints (e.g., GDPR).
4) Deterministic handoff
Package a clean summary (questions, answers, attachments, decision points) and route to the right agent queue. This eliminates screening overhead and speeds up advisory contact.
WhatsApp Insurance Bot Setup: Step-by-Step Implementation
Implementing a policy‑compliant WhatsApp bot is straightforward when you separate channel, logic, and knowledge:
- Channel: set up the WhatsApp Cloud API and verify a business number. Configure webhooks for message events and media.
- Logic: run a backend that handles intents, Q&A, and forms. Use rules for guardrails and an LLM for natural replies, with retrieval from your curated KB.
- Knowledge: maintain a vetted source of truth (plans, eligibility, exclusions, territories). Version it and add review dates.
- Compliance: enforce the 24‑hour service window, use templates for business‑initiated outreach, provide human escalation paths, and log consent.
- Handoff: integrate with your CRM/ticketing for assignment and SLAs; attach the conversation summary and form data.
Insurance Chatbot ROI: Cost vs. Conversion Analysis
ROI depends on the value of additional qualified opportunities and the operating time avoided. It should be measured against a baseline using comparable periods, with changes in demand, campaigns, pricing, and staffing recorded.
Basic model:• Additional qualified leads × close rate × average policy value • minus platform costs and remaining operating time
Before launch, record the qualification rate and average handling time. After launch, compare equivalent periods and annotate seasonality and campaign changes.
ROI model
Insurance Chatbot ROI Calculator
Estimate the time saved handling WhatsApp conversations and the monthly value of that capacity.
The selected amount is an editable model input, not a current FRAI plan or quotation.
Assumes 46 work weeks per year and 40 hours per week. This estimates capacity you can reallocate, not headcount. Actual software scope and pricing require a proposal.
Where AI fits—and where it does not
In this article we covered the HI Expats case study which has a focus on lead generation. The same approach can support other workflows when the task, source data, controls, and human responsibility are defined first.Use-Case Library · 22 of 22
Use-Case Atlas: Insurance Chatbots in Action
FAQ
Frequently Asked Questions
How does an insurance chatbot work?
It answers from a reviewed knowledge base, collects the information required by the defined workflow, and passes the conversation context to a human agent. The agent remains responsible for advice and final insurance decisions.
What did FRAI automate for HI Expats?
FRAI built and operated the initial question-answering, data capture, qualification, routing, and follow-up workflow. HI Expats handled advice, quotes, underwriting, and policy decisions.
What result was measured?
Across two consecutive 92-day periods, completed policy sales increased from 34 to 57, qualified leads from 176 to 288, and average manual qualification time fell from 16 to 10 minutes per qualified lead.
Does the comparison prove that software caused the increase?
No. It is an observational before-and-after comparison, not a controlled experiment. Demand, campaigns, seasonality, prices, staffing, and other commercial conditions may also have influenced the result.
What is needed to operate this kind of workflow?
A reviewed knowledge source, defined routing rules, consent and retention controls, access control, logs, monitoring, documentation, and a clear human escalation path are required.
Services
Recommended services
Curated for this article. Pick one or ask us for something custom. Continuous automated monitoring. Human response coverage is defined in the agreement.
Managed Website Lead-Capture Chatbot
Managed website lead-capture chatbot with qualification, CRM handoff, monitoring, hosting, maintenance, and controlled human follow-up.
See all services →Service · Go-live 2-4 weeksManaged Product Discovery System
Managed marketplace product discovery with search, recommendations, integrations, infrastructure, monitoring, and monthly improvements by FRAI.
See all services →Service · Go-live 2-4 weeksManaged WhatsApp Support System
Managed WhatsApp support with intake, tickets, order checks, human handoff, monitoring, maintenance, and monthly improvements.
See all services →