Fraud Prevention in Banking: How Salesforce and Grazitti Empowered Faster Threat Detection

While we’re securing the present, fraud is racing ahead to the future.

A new study by Juniper Research has revealed something big! Fraudulent banking and money transfer transactions are expected to reach $48 billion globally by 2029.^[i]

That is a 153% increase from the current $19 billion in 2024.^[ii]

Fraud is growing fast.
Banks need faster, smarter ways to protect their customers and their business.

One of our customers, a leading financial institution, decided to move early.
They partnered with us to strengthen their fraud prevention approach.

Together, we created a solution that detects threats faster, reduces risk, and builds customer trust.
A solution that grows stronger with every transaction.

Here is how we helped them build a smarter, safer future.

Know the Customer

The customer is a personal finance company that provides services including free credit scores, credit reports, credit monitoring, and financial recommendations.
They employ customer service agents who interact with members through web forms, email, phone calls, and internal tools.

As a financial technology company handling sensitive personal and financial information, it maintains strict security and compliance requirements to protect member data and the integrity of its systems.

Understanding the Fraud Pattern Observed in Customer Interactions

The banking sector is facing a growing wave of sophisticated fraud attempts.
In this case, external threats were targeting the bank’s customer service agents, prompting unauthorized conversations through platforms such as WhatsApp and Telegram.

Once agents moved to these unmonitored channels, fraudsters could carry out activities without being detected by the bank’s security systems.

The bank identified a clear pattern:
Fraudsters were using specific keywords, such as “WhatsApp” and “Telegram,” across various communication channels to initiate these unauthorized interactions.

Agents who responded to these messages intentionally or unknowingly were at risk of becoming part of the fraudulent schemes.

These fraud attempts were happening across all customer service touchpoints:

• Web-based case submissions
• Email communications
• Phone conversations
• Chat Conversations
• Internal communication tools

Key Business Priorities

The primary goals for this project were clear:

• Detect Potential Fraud: Build a system to identify suspicious communications using predefined keywords across all customer service channels.
• Protect Agent Integrity: Safeguard customer service agents from being compromised by external parties.
• Enable Swift Response: Instantly notify the security team when suspicious activities are detected.
• Create an Audit Trail: Maintain detailed records of all flagged communications for investigation and compliance.
• Stay Adaptable: Design a solution that can quickly evolve as fraud tactics change over time.

Solution Components and Technical Implementation

The solution we offered combines standard Salesforce capabilities with minimal custom development to monitor all communication channels effectively. Key components include:

1. Centralized Keyword Management

• Custom Labels:
  Custom labels were added to identify and flag suspicious keywords. Here’s a breakdown of the labels implemented-
  • “SpammerKeywordEmailRecipients” stores the email addresses of the security team who will receive alerts.
  • “SpammerKeywords” holds the list of suspicious keywords to monitor across all channels.

2. Case-Level Tracking

To ensure case-level tracking, we implemented custom fields- 

• Custom Field:
  • “Spammer_keyword” is added to the Case object to store detected keywords.
  • This enables detailed tracking and reporting of suspicious cases.

3. Multi-Channel Monitoring

To secure communication channels, we implemented apex classes:

• An Apex class was implemented to scan all incoming web case submissions for suspicious keywords. When detected, the case is flagged, and the security team is notified.
• Similar to web submissions, an Apex class was implemented to scan incoming emails (subject and body) for suspicious keywords, triggering alerts to the security team.
• Salesforce Conversation Intelligence (out-of-the-box functionality) was used to transcribe phone calls and analyze the text for suspicious keywords, triggering notifications if necessary.
• A feature that detects keywords in conversations between the agent and the customer was created. With this, when a keyword is found, we send a notification and save the keyword at the case level. This also helps with reporting.
• The same keyword detection logic is applied to data from internal systems, ensuring consistent monitoring across all communication channels.
  

4. Notification System

We made some changes in the notification system and initiated: 

• Email Alerts:
  When suspicious keywords are detected, the Spammer_keyword field is populated on the case, and an email notification is sent to the security team.
  The email includes case details, messaging session numbers (if applicable), and detected keywords.

Impact Analysis and Key Takeaways from the Project

A closer look at the outcomes reveals how the right strategy and technology choices helped the customer drive stronger resilience, operational efficiency, and sustained customer confidence.

Reduced Fraud Incidents:
Early detection of fraudulent solicitation attempts helped the customer prevent the successful compromise of agents, significantly reducing fraud risk.
Faster Response Times:
Automated detection and immediate notifications streamlined the process, ensuring the security team is quickly alerted to suspicious activity, allowing for faster action.
Comprehensive Coverage:
By monitoring all communication channels, including web forms, emails, phone calls, and internal tools, the system eliminated potential blind spots in fraud detection, providing 360-degree protection.
Centralized Reporting:
All suspicious activities are now tracked directly in Salesforce, enabling effective trend analysis and pattern recognition. This centralized approach provided valuable insights for ongoing fraud prevention efforts.

The Bottom Line

According to reports^[iii], the growing sophistication of fraudsters, especially with the use of artificial intelligence (AI), is placing immense pressure on banks. As threats grow more sophisticated, reacting simply won’t cut it. To stay ahead, banks need bold, proactive solutions—ones that not only preserve trust and ensure compliance but also protect future growth. That’s why banks must choose their platforms wisely, opting for solutions that are smart, scalable, and secure enough to withstand the next wave of threats.

If you’re looking to strengthen your fraud prevention efforts, now is the time to act.