AI Phone System for Insurance Agencies: Unipol Cerignola
725 inbound calls managed by AI for Unipol Cerignola Agency: 96.6% successful transfers and 93.2% quality in one month. Want a phone system that never misses a beat?
At an insurance agency, the phone never brings the same request twice in a row. Between quotes, claims, appointments, and simple information, order is essential. In this DeepAgent × Unipol Cerignola Agency case study, we introduce Adriana, an AI phone system for insurance agencies that precisely manages the telephone front office: in two months, it handled 725 calls, completed 200 transfers, and generated 90 opportunities, increasing quality from 63.4% to 93.2%. ## TL;DR - 725 inbound calls managed in 2 months with DeepAgent Adriana. - 200 successful transfers out of 207 attempts (96.6%) to agents. - 90 commercial opportunities: 60 call-backs, 17 quotes, 13 appointments. - Quality of correctly handled requests increased from 63.4% to 93.2% in one month. - 99.6% call transcripts: precise verification and continuous improvement. ## The Client ### Who they are The Unipol Cerignola Agency, in the province of Foggia, is led by Domenico Fratepietro (Unipol General Agent) and has been active as insurers since 1973. A historical presence in the territory, where the phone remains the primary channel for assistance and new requests. ### Starting point One number for all needs. The goal: to make it a reliable and traceable channel, with every call handled, correct transfers when a person is needed, and complete data to process cases, without losing any requests. !AI Phone System for Insurance Agencies: Unipol Cerignola — figure 1 ## Objectives and Challenges Adriana's goal is not to sell, but to manage every request well and guide it to the most useful action for the agency, while measuring the service. Three problems to solve: - Immediately understand the need: different paths for claims, quotes, information, and appointments. - Transfer the call when a person is needed: the transfer must work without leaving the customer on hold. - Measure the service, not the sale: success = verifiable and correct outcome. This is the work expected from a virtual telephone assistant for insurance: welcoming, understanding, directing, without replacing the consultant. ## How the AI phone system works for insurance agencies Adriana's workflow is built in five essential steps: 1. Welcome: answers, identifies the reason for the call, and collects initial information. 2. Classification: directs to six main categories (transfer, quote, call-back, information, claim, appointment). 3. Management or transfer: handles simple requests directly; when a person is needed, it transfers the call to an agency agent. 4. Data collection: acquires necessary data for quotes, appointments, and call-backs; appointments are directly added to the calendar. 5. Quality control: outcomes and transcripts are reviewed to ensure each call was routed correctly. Even telephone management of claims goes through here: Adriana takes charge of the report and routes it to the correct contact person. !AI Phone System for Insurance Agencies: Unipol Cerignola — figure 2 ## Results: 725 calls, 343 operational outcomes Below are the main indicators for the analyzed two-month period. | Indicator | Result | |---|---| | Calls managed in two months | 725 | | Conversation time | approx. 10 hours and 15 minutes | | Average duration | 51 seconds | | Successful transfers to agent | 200 out of 207 attempts (96.6%) | | Interested clients for call-back | 60 | | Information requests handled | 47 | | Quotes collected | 17 | | Appointments created in calendar | 13 | | Claims handled | 6 | | Transcription coverage | 99.6% | Summing the outcomes yields **343 operational actions**. Among these, **90** are commercial opportunities: 60 call-backs, 17 quotes, 13 appointments. The sum is by category: a single call can generate more than one outcome. !AI Phone System for Insurance Agencies: Unipol Cerignola — figure 3 ## Quality that matters (and grows) Let's look only at calls longer than 30 seconds, those with a substantial request. The question is: Did Adriana understand the need and conclude with the right action (agent, quote, claim, appointment, information, call-back)? The answer is yes in **63.4%** of cases in July and **93.2%** in August: **+29.8 points** in one month. With **99.6%** of calls transcribed, every outcome is verifiable. !AI Phone System for Insurance Agencies: Unipol Cerignola — figure 4 ## What this case teaches us - Measure service, not sales: success = verifiable outcome consistent with the request. - Transfers are key: 200 calls out of 725 required a person; ensuring they reach the right person is crucial. - Consolidate quality: 93.2% is the level to maintain with constant reviews of calls over 30 seconds. - Read transcripts: with 99.6% coverage, every error becomes a script correction. ## Related content - AI Phone System for Medical Practices - AI Phone System for Restaurants - AI Voice Agents in Banking and Insurance - Case Study: Insurance Agent Recruitment (outbound B2B) ## Conclusion For the Unipol Cerignola Agency, a well-designed AI phone system means every call goes to the right place: to an agent when needed, to the calendar when it's an appointment, to the quotes list when it's an opportunity. Want to see Adriana in action? Request a demo and try a DeepAgent agent in real-time: leave your contact details, and you will be called back from +39 06 8345191. For the complete case study: giorgio.giovanardi@deepagent.app. ## Frequently Asked Questions ### What systems does the voice agent integrate with? Adriana connects to the existing VoIP phone system and can exchange data with CRMs, calendars, and internal agency tools. Integrations enable automatic actions (appointments, notes, tickets) and end-to-end traceability of the front office. The goal is to have complete data on every call and reduce manual work for the team. ### What happens when a customer asks to speak with a person? When the request requires human support, the voice agent transfers the call to the most suitable operator and monitors the outcome. This flow reduces wait times and transfers, maintaining the conversation context. In this case study, transfers occurred with a 96.6% success rate, improving customer experience and operational continuity. ### Can it manage claim reports? Yes. The agent receives the report, collects essential data, and routes the case to the assigned contact person. The process is structured for speed and completeness, minimizing callbacks and missing information. Case management is tracked, with searchable transcripts for internal audits and continuous improvement of operational procedures. ### How long does activation take and how is success measured? Activation is quick: once the script and integrations are defined, it goes live in a few weeks. Success is measured by operational indicators: call handling rate, successful call transfers, quality of outcomes on substantial calls, data completeness, average conversation time, and correctly routed reports. ### How are privacy and compliance managed? Calls can be recorded and transcribed according to policies agreed upon with the agency. Information notices, consent, and retention times are configurable. Data is processed in compliance with GDPR, with access controls, audit trails, and the possibility of anonymization or deletion upon request. ### What is the impact on an agency’s ROI? By reducing wait times and missed calls, increasing successful transfers, and improving the quality of outcomes, the agent creates efficiency and new revenue. In two months, 90 concrete opportunities (call-backs, quotes, appointments) and 200 correct transfers freed up team time and improved customer satisfaction. Talk to sales → · Create your agent →