Did anyone ever tell you to be creative about something? When this situation happens, I use to close my eyes and start repeating this mantra to myself: “Let’s go creative! And by the way, if you can do it right now… it will be perfect!”
This situation is what is referred to as a “double bind” - a tricky way to communicate two discordant instructions given simultaneously. Another example of a double bind is telling to people to be spontaneous. If they begin to act spontaneously, they are not truly acting spontaneously; they are simply following instructions. In the end, it is a no-win situation.
From psychology to global level, the double bind constantly occurs. For example, a double bind is encountered in economy when countries face the challenge of growing economically without disrupting the environment.
In Telco, a variety of double bind situations can also be presented. One type that is particularly worrisome to operators is the double bind faced in fraud management. In fraud situations, operators set the necessary fraud rules to prevent future fraud cases of the same type. Instead of controlling fraud activity, this actually results in the growth of a constantly adaptive capacity of fraudsters to escape from the fraud management system based on traditional fraud rule mechanisms. But, on the other hand, doing nothing is not an option. This is another no-win situation. As traditional fraud rules increase in number; the more complex and tricky schemes to perpetrate fraud become, thus making it more difficult for operators to reach the level of sophistication that fraudsters are achieving.
Fortunately, advanced analytics are now able to confront the fraud challenge. The capacity to analyse data on a multiple dimension reality, combined with the performance of Big Data processing, makes use of detection and prediction models on large volumes of data more realistic now than it was a decade ago.
Detection models, also called unsupervised models, are mainly used to identify groups or clusters from a customer base so that the behaviour of entities within each group are more alike than the ones belonging to other groups.
When applied in the context of fraud, detection models can be used to do the following:
- Detect suspicious behaviours
- Identify new fraud types
- Enrich case management with cluster information
- Reduce false positives
On the other hand, prediction models analyse past behaviour to obtain the propensity value of a particular event to occur.
Prediction models can be used to prevent the occurrence of fraud. These models take into account the behaviour generated by entities upon their traffic data and evaluate their propensity to commit fraud. The prediction models can be used to do the following:
- Prevent the occurrence of fraud, scoring all customer base and identifying the entities having a behaviour more likely to commit a fraud
- Enrich case management with additional information to better evaluate false positive
- Prioritize the cases in Case Management with additional propensity values information
Advanced analytics are leading to new weapons in the fight against fraud ecosystems that are continuously growing in complexity.
The added value of such techniques is their capacity to rapidly adapt to new realities that are increasingly more difficult to achieve with the setting of traditional fraud rules.
Another trend that we assist is the constant concern of providers in delivering user-friendly advanced analytics tools, solutions, and products that are capable of guiding a fraud analyst through their investigation, from selecting the customer base for analysis to reaching actionable results.
Despite the double bind suffered from traditional fraud rule mechanisms, the use of advanced analytics tools can produce the optimum combination, leading an operator to the next step of a Fraud Management System, introducing more advanced and complex rules within a large volume of data.
With advanced analytics tools, double bind in fraud management won’t be an issue for much longer.
Let Us Know What You Thought about this Post.
Put your Comment Below.