Even without the “e”, KYC (know your customer) for the regulated financial and banking sectors comprises a wide spectrum of activities. And the responsibility of KYC scope expands over time to cope with prevention of financial crimes, especially the AML/CFT/ABC (anti-money laundering/countering financing of terrorism/anti-bribery and corruption) compliance. Prior to assessing a customer’s financial risk, identity verification (IDV) is traditionally the first and an important pillar within KYC process. Australian’s system uses the 100-point check, while Malaysia’s financial system relies on MyKad-fingerprint verification in face-to-face IDV.
And as electronic KYC is becoming the new normal due to evolving landscape in client onboarding trends, most if not all of the KYC activities are to be automated online with straight through processing, such as facial recognition for facial sighting and online API for financial risk assessment. At the same time, deep techs such as artificial intelligence and big data analytics are embraced to unlock new possibilities in financial crime prevention and create new opportunities from financial inclusions.
If we speak to 1 person to understand eKYC, we may only gain a tip of an iceberg. That is why, we gathered 7 speakers in Kuala Lumpur, and 6 speakers in Jakarta, to share their respective views on eKYC.
Microblink Sees Beyond Text RecognitionMicroblink is a R&D company developing proprietary computer vision technology optimized for real-time text recognition locally on a mobile device. Innov8tif is a regional distributor of Microblink in Southeast Asia, and we are using Microblink’s OCR technology as part of EMAS eKYC suite, to recognize ID card details as part of the ID verification process.
Valentina Kusek and Boris Trubic came all the way from Zagreb, and shared with us exciting updates about Microblink’s next generation OCR technology – DeepOCR, and its model trained for bank/credit card scanning. In addition to ID card recognition on mobile, DeepOCR will also be made available to server-side processing as Web API. Seeing beyond text recognition, Microblink is also working on document authentication technology, such as hologram detection to prevent identity fraud.
Cyber EthnographyToday’s consumer behaviors are getting more sophisticated to be generalized by traditional demographics, such as gender, age group, ethnicity, education level, income group and marital status. Dr. Ian Tan, Data Scientist from Priority Dynamics, pointed out a fact:
“In this room, we have 25-year-old and 45-year-old. It shows that there is a common trait between these 2 age groups and they do behave the same in some aspects, or share common interests”Thus, the traditional market research or consumer analytics is no longer entirely valid to derive customer insights and profiling. This leads to Dr. Ian’s sharing on “Cyber Ethnography Support for Risk Mitigation”. The word ethnography, is a new word to most of the event attendees. By definition, cyber ethnography is explained as:
online research method that adapts ethnographic methods to the study of the communities and cultures created through computer-mediated social interactionDr Michal Kosinski, whilst at The Psychometrics Centre, Cambridge University
- Showed that Facebook “likes” reflect how extroverted and intellectual the person is.
- Analysing the words used in tweets, number of tweets and the number of followers they have, can indicate the person’s personality.
- Analyzing tweets can also infer the dark side of a personality, e.g. narcissistic or even a psychopath.
Thus, by applying cyber ethnography methods, coupled with data analytics, a financial company can derive unconventional credit risk assessment using psychometric and behavioral science data. This discussion opens up a wide spectrum of new possibilities for risk and fraud analytics.
MyKad Document AuthenticationNathan, Innov8tif’s Head of Product, shared our latest development on document authentication technologies for MyKad, aimed at helping to automate compliance and prevent ID fraud. For example, document layout test is automated using computer vision technique to ensure that no part of the submitted ID image is blocked or covered, by a finger for example. This doesn’t just optimize customer onboarding experience through instant feedback, but also reduces overhead on compliance review.
What about spotting the fake/replica, or self-printed ID? A customer doing self-registration could have made a fake ID by altering some key ID details such as ID number and full name, and printed the card. Then, took a photo of this card, and submit it online. In this scenario, even the compliance officer examining the ID image from a computer screen, will likely not be able to spot the fake. To tackle this challenge, Nathan shared a document authenticity check – known as microprint test, to differentiate between an image captured from genuine MyKad, vs. an image from a self-printed ID. The devil is in the detail.
Fighting Financial Crime with Big Data“If you think compliance is expensive, try non-compliance”, said former U.S. Deputy Attorney General Paul McNulty.
As part of customer due diligence process, how much you know about your new customer depends on how much intelligence you are presented with from credible sources. While “more is never enough”, it is technically challenging for a business to mesh multiple sources to derive conclusive knowledge about a person. We are talking about data on risk of money laundering, terrorism financing, sanctions, trafficking, smuggling, counterfeiting, slavery and more.
Candice Lai, Senior Account Manager of Refinitiv (formerly Thomson Reuters Financial & Risk), shared with us the human element intelligence of Refinitiv’s World-Check data, sourced by 400+ full time research personnel speaking 60+ languages. Coupled with online API, subscriber of Refinitiv’s data can benefit from real-time straight-through customer due diligence screening.
Accelerating Financial Inclusion with Digital eKYCThe cost of remittance is often measured by monetary cost incurred on the payment processors and customer onboarding channels. However, if we examine further, we would also realize that opportunity cost is incurred on the remittance customer, often the foreign workers, for the time spent on traveling from rural areas to physical counters or branches of remittance service provider.
With eKYC, these migrant workers can be served from anywhere connected via mobile network, from the convenience of a smartphone. Michael Tiew from Cyber Village shared their customer use cases, on how financial inclusion is accelerated through modern technologies. For example, cash remittance is authorized safely and reliably using facial recognition.
Customer’s Feedback from YoodoYoodo, a digital telco from Celcom Axiata, is one of our pioneer customers of eKYC solution. Yoodo’s business operates purely on digital channels with zero physical branch or dealer, and Yoodo has gained the top spot in Social Reputation Score among all mobile network operators, as reported by Synthesio (Feb – 31st Oct 2018).
It was 2016 when we firstly started to explore a tip of the mobile ID verification solution iceberg as our new business vertical, by offering OCR scanning of MyKad. It was the telecommunication industry that we inked our first deal, before the financial services sector started to look out for mobile OCR for ID scanning. The telco industry has always been a key customer segment for us, as it often involves large volume of direct B2B users and B2B2C end-customers. It’s a high-pace segment that demands speed of doing things, and any solution that is put into the wild can also be better testified for product-market fit. In early 2018, Yoodo launched its full-digital products into the maket, and Innov8tif is fortunate to partner with Yoodo to empower the digital KYC solution using facial recognition and ID card scanning.
An interesting incident occurred. We didn’t realize that our product was, unintentionally, “racist”. Facial liveness detection requires that the mobile camera captures decent lighting, or the object isn’t experiencing a back-lighting condition which may lead to a darker face. One of the Yoodo’s customers experienced such situation, and the liveness detection component displayed a message phrased as “Face too dark”. What we actually tried to suggest, was to advise the user to move to a spot with better lighting condition. And the customer wasn’t happy to see this unfriendly message. On the positive side though, the customer reported this incident to Yoodo’s customer care, and we benefited from such feedback to ensure similar incident doesn’t occur in the future. It’s a priceless feedback. Thank you, Mr. Farid Yunus - CEO of Yoodo, for the kind sharing of your experience as a customer of Innov8tif's solution.