Technology adoption and utilization continues to be critical for digital businesses, particularly when it comes to tackling the ever-changing fraud landscape.
As businesses outgrow their in-house developed solutions, or look to add to their current stack, outsourcing fraud detection capabilities is commonplace. Cloud technology empowers increased flexibility and scalability, while at the same time reducing running costs. It also enables access to the latest fraud-fighting capabilities such as digital identity recognition, behavioural biometrics, and machine learning.
However with so many different solutions and buzzwords at large on the market today, how does a fraud department ensure that they are selecting the best technology for their needs? In this short blog we’ll look at some of the key considerations businesses can make before taking the plunge into their next fraud spend:
Do I understand my fraud threats?
Many different fraud typologies existing in online fraud. These include account takeover, automated BOT attacks and scams. A fraud department should clearly understand what is driving their fraud losses by consistently labelling and categorising their fraud data. This creates transparency and enables a business to see exactly what problem is putting the most pressure on their fraud budget. Prospective vendors can then be challenged on how they can address these specific fraud types, giving the fraud department comfort that the chosen solution will have the desired impact once it is live.
See through the buzzwords
Buzzwords terms like ‘Artificial Intelligence’ or ‘Machine Learning’ are likely to appear in most sales pitches. While these terms and functions undoubtedly have their place, businesses must realise that there are certain prerequisites required for technology such as machine learning to work effectively. Models are typically trained to recognise patterns that already exist in data. So for example a smaller institution, with low data volumes and only few confirmed fraud markings, is unlikely to yield good long-term results from machine learning. In this case a consortium / crowd-sourced form of machine learning combined with a rules-based approach might be better suited.
Can other departments benefit from my fraud investment?
Fraud tools often provide useful ancillary functionality for other areas of an organisation. For example Financial Crime Compliance, Cybersecurity and other departments can all potentially benefit from fraud tech such as device fingerprinting and location analysis. Creating collaboration with other business units and involving them in the solution selection process can make building your overall business case easier, as wider business value can be realised from the same investment.
This is two-fold. Firstly, does the solution provider offer personnel that have practical experience in your industry, and are they located geographically near your fraud team (or willing to travel regularly)? Even the best fraud solutions will only succeed if the right people are working with them. Some of the best product and business discussions take place in the moment; therefore having dedicated and engaged people that know your business well will drive maximum long-term value.
Secondly, does your fraud department itself employ the right people. If a fraud team does not have the right level of technical and analytical skills to push the boundaries of a fraud solution, you won’t achieve optimum results. Having the right staff allows you to customize the solution for your specific needs, instead of relying on out-of-the-box standards, therefore improving outcomes. Personnel selection is just as important as solution selection, so don’t overlook it.
Delivery and value timescales
Understanding the delivery timescales of the solution provider, implementation support that they offer, along with what technical resource your department needs to successfully deliver, should be a key part of the decision-making process. A solution that can solve all your problems but takes a year to deliver, may not provide a better 3-year ROI than a solution that only covers 80% of your problems but can be delivered in 3 months. You should also be looking for your fraud solution to be appropriately tailored and to find fraud from day one of go-live, so ask how this can be achieved.
Benefits beyond just detecting fraud
Invariably a fraud department will need to seek budgetary approval to spend money on a new solution. This involves the need to clearly lay out what the anticipated benefits will be. To effectively build this business case, it can be helpful to look beyond just the value of fraud savings. For example:
- Does the fraud solution allow you to reduce false positives, therefore reducing the number of human reviews required?
- Does this improve the overall user experience for your end consumers, and if so, can this be quantified as part of the business case?
- Does your new solution enable you to retire an in-house or incumbent fraud solution, therefore reducing overall department run costs?
This list could go on but thinking outside of the box is the key here. You should also challenge your preferred vendor to support with business case development. They should be able to provide real-life examples of benefits they have recently delivered within your industry.
Closed box vs data transparency
Though becoming more popular, this is still not a standard feature for some fraud solutions. It’s important for fraud departments to understand how a decision was made, so receiving an unexplained outcome from a “closed box” solution is not ideal. Having decision transparency enables solution awareness, stakeholder management, and if desired, customer management during false positives and referral discussions. All data used in the decision should also be available post event for forensics and interrogation.
Data generation for a layered defence
Most fraud systems are capable of consuming data relating to an event and analysing that data in real-time to produce a risk decision. More sophisticated solutions however will also generate their own data, and couple that with the data provided by the business to reach an augmented outcome. This layered approach offers enhanced decisioning power while providing additional assurance and adaptability for changing and emerging fraud threats. Vendors should also be challenged about future roadmaps, as well as release frequency, to support regular tool enhancement.
Finally, do I really need something new?
A basic but essential check. Many businesses are out-of-touch with the latest offerings of their incumbent vendors. Understanding new features in your existing products and combining it with thorough utilisation from a people and analytics perspective, will invariably result in improved fraud detection. Only if you’re confident that you’re maximising your current solutions should a new one be considered. Furthermore, businesses should always be pushing their vendors to continually improve their products. Fraudsters never stand still, fraud technology shouldn’t either. A fraud solution that isn’t evolving is actually moving backwards.
Subscribe today to Fraud and Identity in Focus to receive regular email updates.