Network analytics in action

Examples across all areas of the business.

Anti-Money Laundering (AML)

Transaction Monitoring

If you are going to launder money at scale, then this will be an organised activity involving “controlling minds” and networks. It therefore comes as no surprise that it is common to find networks of bank accounts, customer details and transactions linked together.

What is a concern is when you get lots of alerts clustering around a network making it an outlier. Imagine a network with 6 amber accounts, 1 red flag and 3 green accounts – this is something you could probably escalate to Level 3 AML review automatically. Conversely, one amber account in a network of green accounts is highly likely to be a false positive. Resolving entities to enrich AML alerts and then building networks is one way to speed up Level 1, 2 and 3 reviews.

Know Your Customer (KYC)

By using third party data, such as directors and shareholders, sanctions and adverse lists, networks can be built within your current customer and prospect data. This allows you to see any connections to high-risk factors whether direct or through their beneficial ownership structures.


By linking data together, it is possible to create more context and make better informed decisions. For example, a bank has a small business as a client, it also banks 3 out of the 4 directors. The small business approaches the bank for a loan.

Quantexa technology builds a network and could apply the following basic rules in combination with traditional predictive analytics:

  • The 3 directors have all taken out personal loans recently;
  • The 3 directors have growing balances on their credit cards;
  • By examining payment information, one of the major customers of the small business has ceased making regular payments. This same issue has been seen with this customer for other small business clients of the bank.

Based on this network view combined with traditional predictive credit risk models, it would suggest that further loans to the business may be very high risk.


By enhancing fraud detection models with network analytics and ensuring alerts are suitably embellished with reasons, significant benefits can be unlocked:

  • Up to 70% more fraud is detected, including the complex organised criminality associated with significant reputational damage and losses;
  • False positives can be drastically reduced, driving hit rates up from 1 in 10 to 8 in 10. This in turn reduces wasted effort and makes teams effective;
  • By presenting the user with the full picture of the alert and the context, investigation times can be reduced by 60%+.

Insurance claims fraud example: A claim is presented for a middle-aged driver of an old vehicle, he and his friend are also claiming for neck injuries. Is this fraud? Possibly. Examining his history as an entity, it is the first year with the insurer. However, the network reveals that another person whose policy was purchased with the same credit card also had an accident in the last 3 months, involving a neck injury. Also, the car he crashed into had an accident 6 months ago, involving neck injuries. Fraud? Almost certainly.


Building networks within your data and using 3rd party data provides a better view of your customers and prospects. Here are some examples:

  • Small to medium business prospecting

Using data sources that have details of shareholder and directors of businesses along with basic financial information allows you to build networks of businesses. These may have common shareholders or historic director relationships as well as shared addresses or accountants and legal firms. By combining this insight with your current customer and prospect base you can enhance what you know and who you know within your prospect base. You can target businesses and use your relationships to get introductions.

  • Contagious churn

Organisations often sell customers multiple products, but then find that if they churn off one product, they will also drop others. Even worse is when one customer churns they cause other customers to leave too. As well as banking and insurance, it has been particularly common in mobile operators.

Many operators will provide early handset upgrades to retain customers, spending hundreds of millions a year. Imagine if you could be more sophisticated in deciding upon the recipient. Networks analytics has demonstrated that you can predict key individuals who cause contagious churn. By targeting those individuals, customer churn can be significantly reduced.

You may also be interested in…

Compliance & Anti-Money Laundering

Fraud, Anti-bribery & Corruption


Follow Us

Register for a demo

Please leave this field empty.

I have read and understood the Privacy Policy