Time for the Rise of the Machines? Reflecting on the Risk & Assurance Group's London Conference

Hayley Daniels
Author: Hayley Daniels Senior Risk Management Consultant
Date: 4th April 2017
Categories: Technology, Data, Financial, Telecoms, Revenue Assurance, Fraud protection, Optimus, Neural Technologies, Big Data, Revenue Management, Digital Transformation, Digital Integration

We are living in an increasingly digital world, breeding increasingly sophisticated fraud and fundamentally altering the way in which we must tackle it. Using data as a tool for communication is becoming a way of life for the younger generation, with over 60% of under 35’s[1] using instant messaging to keep in touch and over 80% visiting social media sites weekly or more frequently. This presents not only challenges but opportunities for telcos who, given this wealth of social media data, can begin to build Digital Risk Profiles for individuals and use this data for fraud prevention.

By using machine learning techniques to mine this abundance of data, operators can use systems to profile behaviour, looking for patterns across multiple data sets and identifying anomalies. Whereas traditionally, Fraud Managers have used rule-based approaches and looked for unusual patterns across voice-based data, fraudsters have become wise to this and begun to circumvent the system, changing their characteristics to such an extent that they can fool the usual methods of detection. If, however, companies apply machine learning to identify behavioural patterns across wider data sets, accessing data packets and examining their contents, seemingly random behaviours can be identified as fraudulent and then analysed against other consumers exhibiting the same characteristics. This kind of detailed analysis will not only allow companies to detect fraud, but also to understand in the longer term where they are losing revenue.

Though, if machines are the way forward in data analytics and crime prevention, what does this mean for today's analysts currently working in fraud management? Whilst many currently focus on voice and CDR – the data we have come to be most familiar with – the majority of the industry requires more technical knowledge about the fields within data packets to understand what is available and what needs to be examined further. The instant answer is that there is re-training required – as an industry I feel we have fallen behind in this area and perhaps buried our collective head in the sand too long. Now is the time to get set and start the sprint to catch up.

The obvious way forward is to automate systems and processes to free up resource and provide coherent dashboard summaries, including digital identities through social media, IoT, mobile money and so forth. An incredible 46% of budget is still spent on manual reviews. This need for automated systems needs to integrate end to end with every platform across the company and provide data mining techniques such as link analysis, predictive modelling and real-time blocking. By combining relatively simple building blocks such as CDR records or event records with SIP records and routing traffic, operators could in theory create real time blocking, stopping frauds in their tracks. The same goes for mobile data – using VOLTE SIP data for monitoring roaming.

It became obvious to me during the RAG Conference that we are beginning to turn a page as an industry. Whereas in years past it sometimes felt like we were repeating old conversations, there was definitely a realisation that we cannot look at fraud and risk management in the same way that we have done in the past. We are now in a world where the activity is hidden and must be mined out of the data that is ever present and ever increasing. We need to make the changes now and begin to enforce new methods. How quickly this is done is really up to individual companies, and a business case must be made in order to get such changes signed off. This will be hard unless the industry begins to share its losses and case studies more openly, shining a spotlight on frauds old and new, and how they are occurring and evolving. RAG is committed to providing a safe forum in which to share knowledge and is dedicated to assisting its members in the fight against fraud.

Hayley Daniels is one of Neural Technologies' fraud industry experts having spent 22 years working in fraud and risk management.  She is currently on the committee of the Risk & Assurance Group (RAG) and reflects on the recent RAG conference and the future of fraud management.