AI Agent Revitalizes Dormant Finance Customers: Case Study
A "cold" database of former customers starts generating concrete conversations and reactivations thanks to an AI voice agent trained on specific context. Want to see how?
You have a database full of people who paid in the past but no longer respond. They are not cold leads; they are former customers. Emails go to promotions, SMS messages are ignored. In 35 days, we tasked an AI agent for reactivating dormant customers with a list of 20,000 former students from a leading financial training brand: 8,805 calls, 1,770 reactivated customers, a conversion rate of 20.1% from an archive the company considered lost. ## TL;DR - 20,000 former students contacted with a natural voice trained on the brand. - 8,805 calls, 1,770 confirmed reactivations, 20.1% average conversion. - Automatic recall every 18 hours for non-responders, without excessive pressure. - Cost of €0.50/min, complete transcripts, and data updates in CRM. - Zero operational burden for the team: focus on high-value follow-ups. ## The Client: 20,000 Dormant Former Students in an Archive The protagonist is a leading Italian financial training brand, specializing in economic education and investment courses for individuals and professionals. Over the years, it has built a large and diverse customer base: advanced students, participants in single events, occasional readers. At the time of the campaign, 20,000 contacts were dormant: people who had previously paid but no longer engaged.  Declared objective (August 2025): to bring those people back into a training event. The project is internally named “Wake Up Call”. ## The Challenge: When Emails and SMS No Longer Land - Saturated digital channels: near-zero response rates for emails and SMS. - Unscalable manual management: impossible to call 20,000 former customers with personalized conversations. - Outdated data: unverified contact information, no recent profiling. - Diverse base: different messages for different profiles, requiring personalization. ## Why a Voice Agent, Not Just Another Email The choice fell on “Alba”, an AI voice agent, to reconnect in a human and personalized way with tens of thousands of contacts using a lean team's resources. A warm, slightly regional female voice, an empathetic and energetic tone, 9:00 AM – 9:00 PM coverage, training on proprietary materials and real objections, recognition of the customer's historical context. This is true conversational phone marketing: every call reopens a relationship and generates useful data. ## How an AI Agent Reactivates Dormant Customers The implementation followed four phases in just a few weeks. **Phase 1 — Technical setup and voice personalization.** Definition of the voice, tone calibration for former students, programming of the 9:00 AM – 9:00 PM window. **Phase 2 — AI training on client materials.** Content integration, uploading brochures and training materials, objections library, and “Wake Up Call” logic. **Phase 3 — Call initiation and data collection.** Massive outreach to 20,000 contacts: reconfirming event attendance, verifying contact details, logging outcomes in CRM via Google Sheet integration. **Phase 4 — Monitoring, iterations, and KPI analysis.** Qualitative feedback, script optimization, objection handling, continuous reporting. A decisive plus: automatic recall every 18 hours for non-responders, without seeming intrusive.  ## The Results: 8,805 Calls, 1,770 Customers Reactivated | Metric | Value | |---|---| | Campaign Period | August 2025 · 35 days | | Dormant Contacts in List | 20,000 | | Calls Made | 8,805 | | Customers Reactivated and Confirmed | **1,770** | | Conversion Rate on Calls | **20.1%** | | Average Cost per Minute | € 0.50 | | Automatic Recall Frequency | every 18 hours | | Conversation Transcriptions and Summaries | 100% | | Reduction of Manual Workload for Internal Team | 100% | The number that matters is 20.1% on a database considered lost: one reactivated contact for every five calls. Quality remains high: positive feedback frequently includes the phrase — “I like it very much” — referring to the calling experience.  ## Costs and Quality: Why It Pays Off The average cost of €0.50/minute changes the math for a customer database reactivation campaign. With a traditional call center, calling twenty thousand former customers with personalized conversations is unsustainable for a cold list. Additionally, 100% of conversations are transcribed and summarized in Google Sheets: verified contact information, recurring objections, real interest levels. Zero distraction for the sales team, who can focus on higher-value follow-ups. ## Replicable Factors for Any Stagnant Database Five elements make the model transferable beyond financial training: - Voice personalization based on the interlocutor's real profile. - Extended time window until 9:00 PM to maximize reachability. - Vertical training on proprietary materials and real objections. - Intelligent recall at wide intervals to avoid fatiguing the contact. - Continuous iteration on scripts and conversational logic. This approach also works for outbound call automation in financial training, but it is applicable to sectors with underutilized historical customer bases. For similar but not identical cases: see the case on never-activated SaaS users [AI voice agent for SaaS user reactivation](https://deepagent.app/agente-vocale-ai-riattivazione-utenti-caso-studio/) and outbound lead pre-qualification for campaigns in the [Fincontinuo](https://deepagent.app/fincontinuo-prequalifica-outbound-caso-studio/) case. ## Conclusion In 35 days, 8,805 calls reactivated 1,770 customers from a base the brand no longer considered an asset. An AI agent does not replace a salesperson: it returns a valuable asset already owned by the company, unreachable through traditional channels. Do you want to test it on your database? Request a demo: you will receive a real-time call from the agent for a practical test. ## Frequently Asked Questions ### How long does it take to start a similar project? From 2 to 4 weeks for setup, AI voice agent training, and basic integrations (e.g., Google Sheet or CRM). The pilot phase allows for measuring critical KPIs (reach, conversation rate, conversion) and optimizing objection handling before scaling up. Times and effort are lower than building a temporary SDR team. ### What CRM integrations are possible? The agent writes outcomes, transcripts, summaries, and updated contact fields. It's possible to integrate common CRMs via API or connectors (HubSpot, Salesforce, Pipedrive) in addition to Google Sheet. Standard fields are defined for reporting and future segmentation, enabling data-driven conversational phone marketing and targeted follow-ups by the human team. ### How is conversation quality controlled? A set of rules, tones, and policies is established, plus a list of approved objections and responses. All calls are transcribed and summarized; logs are sampled for audits, and KPIs such as average duration, outcomes, and sentiment are reviewed. Weekly iterations on scripts and prompts maintain brand voice consistency and increase call conversion. ### What costs should I expect at scale? Operational cost is primarily consumption-based (minutes). With €0.50/min and typical durations of 1–3 minutes, the CPA for former customers is competitive compared to call centers and paid media. The customer database reactivation campaign also benefits from the value of updated data, which fuels subsequent campaigns with better rates. ### Is this approach suitable for very “cold” databases? Yes, because the agent recognizes historical context and adopts an empathetic tone. For older lists, soft recalls and an extended time window work well. Outbound call automation combines with personalization, segmentation, and continuous improvement: often the first conversion is a confirmation of interest and data update.