The fraud tool landscape is changing. Fraud managers are looking for more and more ways to optimise their operations, as the rise in digital payments increases the strain on many anti-fraud systems. Fraud orchestration is set to be the next step in this evolving area. Helping fraud managers to manage fraud problems, both new and existing, without the manual effort that the process usually requires.
Orchestration is the term typically applied to these operations. Where full systems and supporting components can be deployed automatically, making the process very simple, repeatable, and problem free for the operators. Following a core deployment, orchestration can be used for efficiently updating various components, without any downtime. Again, with little or no manual effort.
Cloud services rely completely on orchestration tools, without them, today’s cloud would look drastically different. Orchestration tools allow systems to be deployed, updated, and scaled according to load, such as increasing capacity in a retail payment system during Black Friday sales. The ability to automatically scale is incredibly important to the operator because the cost of operation increases as the size of the deployment is increased. Orchestration tools ensure that only required functions are in operation (often referred to as being ‘elastic’). Optimising running costs, while maximising service.
There are many orchestration tools available. But one of the core concepts behind orchestration is the ability to define deployments in code. Making each deployment completely repeatable, reducing, or eliminating the number of surprises engineers must deal with. One such tool is Docker, which allows DevOps teams to do this with incredible control. Each specific system component (or Docker Container) can be individually specified and controlled, removing any possible conflicts.
Orchestration for deployment is viewed as a mature technology and is relatively sophisticated, however, the adoption of orchestration concepts is only just being explored in many other areas. For example, orchestration is beginning to become prevalent with some machine learning building tools. There are several tools available now, which allow the user to log in, upload some data, train a model, then understand the performance results. The user can then download the model and use it in their system. Examples of these systems include Microsoft’s Azure ML, Amazon’s AWS AI and ML and Google’s AI Hub, to name a few.
These services are excellent tools for many problems, but often require the user to tweak the model and retrain it (model refinement) many times before producing a model suitable for production. There are services emerging which make deployment of these models easier, by incorporating API’s for seamless model updates, however most of the process is still largely manual.
Orchestration in fraud prevention systems
The fraud industry has had access to similar tools for a few years. These tools automate the creation and evaluation of new models, but require manual input to verify performance, and to understand how the new model will affect the fraud strategy performance once it is deployed. Fraud managers want to discover if deploying the model will mean they can retire some problematic rules, or other models due to overlap.
Orchestration concepts can be utilised to answer these questions in an automated fashion. But it can also be made to do more. Indeed, my team and I are using the tools that devops engineers use every day and are applying them to fraud management, and the use case is exceptionally powerful.
The process of developing new models and deploying them in production can be done automatically. Utilising the traditional methods of orchestration to ensure the models are produced quickly and efficiently. This process can also be further improved with the addition of monitoring tools – an essential part of the orchestration toolbox – to understand which models are no longer providing the benefit they once did, and automatically creating a replacement.
Orchestration will not stop there. Complete fraud management will also be possible by extending the classic monitoring tools to focus on maintaining fraud strategy effectiveness. When their performance drops, or new fraud patterns are identified, new models can be developed and tested automatically and optimised into the strategy.
This will be particularly effective for reducing the impact of fast-moving types of fraud. As the system will rapidly update automatically to keep on top of the moving trend – much faster than would be possible if it was left to be managed manually – as well as detecting more fraud and increasing acceptance levels of bonafide transactions.
It is not just the fraud industry where orchestration tools will provide innovation, the concepts will be utilised in many industries for a wide range of problems and business tasks, particularly where staying on top of data trends and reacting to changes quickly is critical. We will see increased automation and optimisation leading to more creative products, services, and staff, thereby creating happier, satisfied customers.
Oliver Tearle is an Innovation Thought Leader at The ai Corporation (ai). ai is trusted around the world for developing innovative technology that allows our customers to create predictable success and grow profitably. Founded in 1998, we have a long track record of providing solutions to some of the world’s largest financial/payment institutions and international merchants. Our longstanding business partnerships are based on making things simple and explainable, both technically and commercially. By focusing in this way, we constantly strive to help our customers create highly profitable returns.