Fraud forecast

What is the challenge?

All companies can be hit by fraudulent acts (such as credit card fraud, phishing emails or fake accounts). Customers of the company (first-party fraud) or external third parties (third-party fraud) can commit the fraud, for example through identity theft. Optimal protection against fraud is relevant – also with regard to international business relationships and different jurisdictions – but also complex. It is often difficult for companies to identify and prevent fraud at an early stage. 

What data can help?

  • Transaction data (master data, remittance data)
  • Image data (e.g. photos of damage)
© Shutterstock
Use case category: fraud forecast

How can companies use their data?

The aim of data-based fraud detection is to detect fraud early. For fraud detection, current data, such as remittance or image data, is compared to data or patterns of fraud. If similar patterns can be found, there is an increased likelihood that fraud has occurred. AI algorithms are also used for pattern recognition. Components include data-based detection of anomalies and fraud patterns to estimate probabilities and block suspicious financial transactions. Machine learning can be used to automatically adapt to new fraud patterns.

Where is this use of data already being applied?

One of Commerzbank’s departments, “Fraud Detection,” develops fraud classification methods and performs investigative data analysis to detect fraud, among other things.

Vacasa, a full-service vacation rental company, is using the service developed by Amazon Web Services to make the booking process more secure. This identifies potentially fraudulent online activity and the creation of fake accounts.

Royal Bank of Canada has a specialized team for data-based fraud prevention. Using social network analytics, data on social relationships between bank customers is collected and stored in graph databases. This helps identify and prevent suspicious transactions.

The start-up RxAll detects counterfeits and defects in medicines with data analyses. For this purpose, data records from drug databases (target values) are compared with actually recorded data (actual values) to determine whether the preparation really contains the active ingredient or is a counterfeit. 

How does this use of data contribute to value creation?

Fraud detection should minimize the vulnerability of company processes to fraud. This makes the value chain more robust. It is also possible to offer new services with the help of fraud detection services and thus generate new value creation with the company’s own revenues.

Aim of data use

Sources: Management Circle AG (2017), Amazon Web Services (2019), Bell, P.C. & Chandrasekhar, R., Harvard Business Review (2017), RxAll (2020)