Intelligent Automation

KYC is the new ​backbone of customer experience in retail banking

KYC backbone of CX
Share on facebook
Facebook
Share on twitter
Twitter
Share on linkedin
LinkedIn

In recent years several high profile fraud cases have stunned regulators and eroded the public’s trust in banking institutions. Take the case of Danske Bank, a large Scandinavian bank involved in an 8-year international money laundering scheme that overlooked €200 billion in payments flowing through the non-resident portfolio of the group’s Estonian branch. This begs the question as to why know your customer (KYC) processes failed to prevent such an oversight.

KYC is a precautionary measure taken by regulated firms to prevent fraudulent activities and has become an integral component of banks’ fraud prevention efforts. It is mandated and essential for confirming the identities of customers during onboarding and throughout their ‘customer lifetime’, as well as verifying their suitability and any financial crime risks they might pose.

A study commissioned by the European Parliament claims that fraud has cost the EU up to €990 billion a year in losses to gross domestic product. In response, stringent regulations to tackle tax evasion and corruption have meant that KYC’s remit has been extended to include continuous monitoring, fraud management, sanctions management and anti-money laundering.

Complying with KYC obligations has always been the responsibility of banks, where KYC capabilities have been developed reactively to meet regulatory mandates. This approach has led to cumbersome processes, fragmented not only across divisional silos but also functional silos within divisions. This is not only an unsustainable model, it puts the burden on the most important and fragile elements; your customers and the customer experience (CX).

How poor KYC can directly impact the customer experience

KYC onboarding is the crucial first step a bank goes through when acquiring lifetime clients. It consists of multiple touchpoints involving various departments, such as operations, legal and compliance. A poor KYC experience has been found to directly affect the customer experience, with adverse effects on your bank’s bottom line. The following are some ways CX is directly impacted by inept KYC. Repetitive and unnecessary requests for information

KYC checks can be invasive during customer onboarding, with banks asking customers for a multitude of documents and personal information (mandated by AML regulations) to build an accurate customer profile. According to a study conducted by Forrester Consulting, customers were contacted on average 10 times during the onboarding process and asked to submit between five and up to 100 documents.

Not only does a poor onboarding experience frustrate your customers, but many people are concerned about how their data is being used, collected and stored. In a Cisco-led survey, it was found that 43% of consumers do not believe they can adequately protect their data.

To eliminate such a problematic hurdle during KYC onboarding, challenger FinTech banks (such as Revolut and Starling) are surpassing traditional retail banks by reducing the number of touchpoints during the onboarding process.

Opening an accountSource: https://builtformars.co.uk/banks/opening/

 

Rising costs of manpower

Not surprisingly, McKinsey stated that “in the United States, anti–money laundering (AML) compliance staff have increased up to tenfold at major banks over the past five years or so.”

Banks are adding staff to vulnerable areas that are in turn causing disjointed KYC efforts across departments, within a single institution. This can mean that customers will be periodically contacted for the same information to fulfil KYC requirements multiple times—leading to duplicated efforts among banking staff and a recurring nightmare for customers.

The kind of acceptable information can also vary depending on the country and the bank, with some institutions requiring a face-to-face meeting to fulfil KYC obligations. This can be a burden on clients who operate across different countries.

Fragmented storage of customer data 

Customer information can also be stored in different systems, departments and even branches. How do we know banks’ KYC records are accurate, if customer data are stored and periodically refreshed in silos that do not seamlessly interact?

This can increase the chances of fraud. If you lack the infrastructure to form an accurate picture of your customer, to begin with, how will you know if you’re letting in a fraudster? Fraud not only causes reputational damage to banks, affecting revenue and growth prospects, but it can also have a direct impact on your customers’ lives. It can affect a victim’s credit rating or result in debt (due to stolen money).

Binary decision-making breeds inaccuracy

Banks currently using linear, rules-based KYC/AML automation systems may be doing more harm than good, as well as putting themselves and their customers at a higher risk of fraud. Such systems have been found to generate up to 90 percent false positives with vulnerabilities that can be exploited through approaches like ‘smurfing’, due to the simple nature of the rules used.

Ongoing KYC as a replacement 

The latest EU Anti-Money Laundering Directive (AMLD5) has made it a requirement for KYC to be an ongoing activity. This means some banks cannot accept certain types of customers, as performing KYC on them would be too expensive. This means closing the door on potential revenue.

Concurrently, ongoing KYC is likely to further inflate compliance costs. A study by Bain & Co. estimates that risk, governance and compliance costs account for 15-20% of the total “run the bank” cost base, among major banks.

With compliance costs becoming so high, banks may struggle to perform day-to-day functions and customer-focussed investments may be deprioritised. But this pattern doesn’t need to be the one that plays out.

Making ongoing KYC the backbone of CX

As we move towards an era of open banking, KYC onboarding will be managed within a connected, intelligent decision automation ecosystem. Intelligent decision automation refers to machines performing “thinking tasks”, which would otherwise require human intervention, while retaining the ability to explain the rationale underlying automated decisions (just as a human would). An example would be a machine deciding which customers meet multifaceted KYC requirements, then repeating this assessment as there are changes in customer profiles and available data. Such a system would be the “brain” into which all other systems plug, so you can get that unified view banks want and provide the seamless experience customers crave.

Intelligent automation, such as Rainbird, sits within the artificial intelligence (AI) space and takes a “human down”, not “data up”, approach to automation. That means we start with human knowledge and apply it to data (so that humans can always understand and explain what the machines are doing). All the logic, ambiguity and rationality behind a specialised human-made decision is combined with data to automate complex human decisions, at scale.

You will be able to dramatically speed up the process of customer onboarding, as intelligent automation can work across siloed divisions, while taking many factors into account and weighing them in a nuanced and efficient manner. False positives can also be reduced, therefore minimising the need for manual investigation.

Intelligent automation systems are also fully transparent, allowing banks and their customers to find out exactly how and why decisions have been made, and which data points were considered. This will instil greater trust in your customers, providing them with complete transparency into decisions or recommendations made about financial products. This can also reduce compliance costs.

The accuracy of customer data is key to ensuring efficient KYC. Where your KYC analysts may be prone to human error or customer data is outdated (due to information silos), intelligent automation software can make decisions despite uncertainty and missing data—it can even gather new information to update erroneous data. Its ability to work with uncertainty will also mitigate potential fraud, by identifying high-risk customers early in the onboarding cycle.

Download our free eBook to find out how intelligent automation can make ongoing KYC a success in your organisation.

Managing continual KYC at scale, using decision automation
How to stop siloed technologies from turning ongoing KYC into a never-ending nightmare.

Join the Newsletter

New Rainbird intelligent automation: if you can click, you can use it

Rainbird For Automation Lovers
Share on facebook
Facebook
Share on twitter
Twitter
Share on linkedin
LinkedIn

Ever worked with someone so good at their job, you wished you could clone them? The thought process goes something like:

“If only every fraud investigator were as good as Taylor. Every problem she encounters, she can solve quickly. Amid a sea of information, she has an eye for pinpointing what’s critical, and what’s utterly irrelevant. Her work is always 30%-40% better than her colleagues’. And she rarely makes mistakes.”

Taylor might seem powered by magic… but she’s not. She’s powered by an excellent mental model of her domain and maths that’s so complex, it comes across as intuition. What’s happening is that Taylor’s mental model, and the maths underlying it, is better than everybody else’s. It’s an abstract phenomenon with material implications. 

Rainbird is a software platform that gives you the tools to copy Taylor’s (or anyone else’s) abstract know-how and represent it visually (kind of like a mind map). That way, a computer can take that knowledge and use it to make decisions in the same domain as Taylor—except 100 times faster and 25% more accurately than her. That is what we call intelligent automation. So, you can have 2 Taylors. Or 100. Or 1000. It’s really up to you. All of them being instructed by Taylor herself.

For the new and improved Rainbird Studio, we set out to accomplish two objectives: 

  1. Make it easier for anyone (really, anyone) to replicate human expertise in machines (we refer to this as building a knowledge map)
  2. Ensure the process of doing so is pleasant

 

No code = instant pro

You don’t need to learn the coding language (RBLang) that Rainbird’s engine uses to understand the visual representations of human knowledge. While you could always build a knowledge map in the visual modeller, being able to code gave some users an advantage. Not anymore. Users can now intuitively apply the majority of key features within the visual interface.

When you want to build representations of your knowledge in Rainbird, you do so using concepts (i.e. a kind of thing) and relationships (i.e. how that type of thing relates to other types of things). For example, “Person” might be a concept, “Destination” might be another concept and “Visits” might be a relationship between the two concepts.

 

Rainbird Studio Canvas

 

So, a “person” “visits” a “destination”—which would help Rainbird understand how people interact with destinations. Every time it then deals with a specific person (e.g. Mike), it knows that it is possible for them to visit destinations (given certain conditions) and can make decisions on that basis. You can then build many relationships and concepts (and more layers of logic and nuance between them all). We’ve moved those deeper layers of modelling right into the visual interface. 

 

A lighter cognitive load

Cognitive overload is a thing of the past with our restructured layout. Our UI overhaul has divided major features into just three sections: 

  1. Edit: where you build models or representations of your knowledge (that look like mind maps) 
  2. Test: where you check that your model is making high-quality decisions, as expected
  3. Publish: where you unleash your model to make decisions in the real world

This way, the structure and hierarchy of Rainbird features make intuitive sense, rather than hitting you all at once. 

 

Neither a decision tree nor a random forest be

We are often asked, “Is Rainbird machine learning?” Sometimes, Rainbird is even confused with decision trees. It is neither—and it’s important to explain why. 

Machine learning refers to systems that can act to give a desired output without being explicitly programmed to do so. A random forest classifier is an example of a machine learning approach that can find patterns in data. And robotic process automation (RPA) refers to the process of programming a computer to take certain actions based on rules expressed as “If this…”, “…then do…” (typically known as decision trees). For example, if “Mike doesn’t have COVID-19 symptoms”, then “he can visit Portugal”.

But these approaches have their challenges. With machine learning, it is difficult for humans to stay fully in control because algorithms find their own patterns in data—and humans don’t always know how or why machines find the patterns they do. So, in sensitive situations—such as credit decisions for minorities or women—we can’t explain why the machine decided one person should get a better credit rate than another. The process is “data-up” (i.e. machines find patterns in data, then impose these on humans), rather than “human-down” (i.e. humans create patterns for machines, which then impose these on data).

With RPA, the decision making process is too linear and unidimensional. You can’t build holistic representations of knowledge—for example, you can’t build a detailed map of concepts and relationships that include COVID-19 risk, job status and destination infection rate. Such that anyone in any situation can consult with the same knowledge map to find out if they can go on holiday. You can only programme one-off steps, for individual scenarios. So, it’s very difficult to build decision trees that can juggle many variables at a massive scale. 

 

 

And that’s why we’ve focussed on two principles as we improve Rainbird: 

  1. Human-down structure: that is, we start with human knowledge and apply it to data (so that humans can always understand and explain what the machines are doing)
  2. Non-linear automation: that is, we focus on capturing and codifying knowledge that can apply to multiple scenarios (rather than on steps to be followed in a one-off, isolated situations)

That’s why it’s important that anyone should be able to use Rainbird and our focus is on knowledge representation (not simply the automation of steps). If you’d like to see how the new Rainbird can support your automation agenda, just request a demo.

Introduction to the new Rainbird
This webinar runs through major upgrades to the Rainbird platform, as well as how they enhance the user experience and provide easier access to key features—without coding.

Join the Newsletter

The future of fraud detection and prevention, in a post-COVID world

Fraud prevention
Share on facebook
Facebook
Share on twitter
Twitter
Share on linkedin
LinkedIn

The global pandemic has brought the world, and businesses within it, to their knees. While communities band together to support one another, not everyone is working cohesively for the common good. 

Cybercrime has spiked since the novel coronavirus reached pandemic status earlier in 2020, with Checkphish reporting a rise from 400 attacks in February 2020 to almost a quarter of a million in May 2020.

Technology has revolutionised fraud detection departments, adding the ability to handle the scale and sophistication of contemporary attacks, but many approaches to fraud detection still fall flat when it comes down to the high level of accuracy required.

But it is the organisation, not the technology, that will be held legally accountable for the decisions made. Choosing a solution that will deliver a high level of accountability and accuracy in fraud judgements is key. Let’s explore how current approaches to fraud risk management measure up.

The conventional approach

It has become common to adopt basic rules-based approaches to fraud risk management. Basic rules-based systems apply a logical program of sequential parameters. They use these parameters to identify instances of fraud and subsequently perform automated actions.

While a conventional approach in fighting fraud, basic rules come with some frustrating pitfalls (although these are circumvented by intelligent rules-based systems, which are covered below). Basic rules-based systems can be expensive and too rigid to scale instantaneously (when most needed). 

The recent surge in advanced fraud cases means basic rules-based systems will struggle to keep pace with the latest techniques by opportunistic fraudsters. More sophisticated fraud solutions are being adopted, as illustrated in a research piece by the Association of Certified Fraud Examiners & SAS, which projected a 200% increase in the use of artificial intelligence (AI) and machine learning to detect fraud.

The adaptive approach 

Machine learning can detect sophisticated fraudulent activities across large volumes of data, ‘learning’ to adapt to new, unforeseen threats, while significantly reducing false positives.

Machine learning programs can be classified into two approaches; supervised and unsupervised.

In the supervised machine learning algorithm, you will present it with both fraudulent and non-fraudulent records. This ‘labelled’ data will help it achieve the level of accuracy to produce a data model, which will then predict whether fraud is present when assessing new, unknown fraud cases.

However, supervised learning is unable to scale on-demand to evolving threats. It requires clean data, substantial computation time for training, and a team of data scientists to build, maintain and interpret.  

Unsupervised machine learning, on the other hand, is useful for spotting fraudulent patterns where tagged data is not available. This model will work off an unlabelled dataset, with unknown output values mapped with the input. It will create a function that describes the structure of the data and flags anything that doesn’t fit as an anomaly. This ultimately saves time and allows the algorithm to mine at scale without the need for human input.

A major drawback to unsupervised machine learning is its inability to accurately explain its results, due to the input data being unlabelled and unknown. This makes it difficult to adhere to compliance measures or explain why a false positive occurred.

The intelligent approach

Intelligent automation (IA) copies (via a visual model) the knowledge of your fraud experts and applies this knowledge to thousands of cases, while instantaneously providing an audit trail for every decision. This includes its certainty level and every factor that went into making each decision, as well as data sources that were accessed. It also includes all considered variables and the quantitative impact of each, dramatically reducing false positives by omitting bias or inaccuracies in initial data collection. 

In contrast to simple rules-based fraud identification tools (like decision trees), the modelling process in intelligent automation allows its decision-making to reflect the ‘greyness’ inherent in fraud, resulting in more reliable outcomes.

It can also handle uncertainties in the dataset. When data is missing or uncertain, intelligent automation will try to find other data to help. If it can’t find supporting data, it is still able to make an inference, which will be presented with reduced certainty. This means that, unlike a fixed rules engine, intelligent automation doesn’t hit a dead end when data isn’t available, or end users are not sure of their answers.

Not only does this system impressively manage scale, deliver a high level of transparency and retain a human-down focus towards data input, the ROI potential of this approach is also a no-brainer. Rainbird has estimated that incorporating IA decision-making into fraud prevention could save UK businesses £7 billion over five years.

What will fraud detection look like post-COVID?

Ultimately, fraud detection and prevention systems that can deliver a high level of accuracy will: 

  1. Reason like human experts do, 
  2. Scale to deal with surges in threats, 
  3. Provide evidence for each decision made 
  4. Be able to handle uncertainty 

When all these elements are present, you will possess a fraud risk management system that will greatly reduce user friction (caused by false positives) and keep you compliant (thanks to complete transparency). 

Unfortunately, COVID-19 won’t be the last widespread crisis we will have to deal with. With climate change and the risk of future pandemics looming, choosing the right approach is not a matter of preference.

To learn more about how intelligent automation fights fraud, watch our on-demand webinar

Free webinar: How to keep up with a changing fraud landscape
Increase fraud detection rates and reduce false positives. This webinar shows how to build an automated fraud system with human intelligence at its core.

Join the Newsletter