The advent of high-speed internet and smartphone penetration that is inching towards the 3 billion mark the world over is making people reclusive. In other words, personal interactions are considered a waste of time, posing a huge challenge for the banking, financial services and insurance (BFSI) industries when it comes to document verification. In this context – “Is the physical presence of the customer really necessary while conducting document verification?”, “Can businesses do with non-documentary verification?” – are the two highly relevant questions before us.
The Challenge: Matching Personal Attributes to the Physical Person
At the core of it, identity is a claim by the user that he or she is a specific person. Of course, the claim cannot be taken at face value—no pun intended. It must be backed by matching of permanent unique identifiers like name, date of birth, biometric data, ethnicity, etc. verification of this identity, on the other hand, seeks to connect these identifiers with the physical person.
In the traditional setup, customers were asked to be physically present to conduct this verification. However, with the arrival of digital banking, this is changing quickly. Customers are no longer willing to be physically present. Of course, with the smartphones handy, it is no longer a challenge. The customer simply has to upload a photograph along with a government-issued ID. Better still, one can upload a video as proof of life and self. Today’s technological advancements even allow for achieving this in real time. However, it brings another problem to the surface. The need for a human being on the other side to verify the identity.
The Problem of Plenty – Solved by Automation
Imagine a popular bank that opens hundreds of accounts each day. It simply cannot afford a workforce that is solely dedicated to verifying these digital photographs and videos. Enter automation powered by artificial intelligence and the problem ceases to exist. Here’s how it typically works: the automated system leads the user through a step-by-step approach, while simultaneously conducting the verification process in the background where the physical attributes are matched with myriad approved identities. This is, in fact, more effective than manual review as the software can spot forgeries with better accuracy, taking into account mathematical comparison techniques. If need be, a layer of manual document validation can still be added.
Here are some of the parameters that an automation solution factors in:
- Cross-document data consistency such as full name, document number, date of birth, etc.
- Signs of manipulation or forgery such as alteration to the original image
- A 3D view of the document, verifying holograms, as they are usually hard to alter
- Document edges that signify superimposition for manipulation
- Overall quality of the document such as optically verifiable ink, text overlay, watermarks, etc.
What Machine Learning Brings to Document Verification
The efficacy of a machine learning process depends on the amount of data fed to it. It uses this data to train itself and learn continually, compounding its performance. Here are some of its salient features:
- Errors owing to fatigue are never a possibility
- The process can catch sophisticated frauds that are not apparent to human beings
- Quick access to global identity documents
- It empowers businesses to offer better service to its customers
- Seamless conformation to global regulations through slight tweaking in rulesets
- Low associated costs
- Instant scalability to any extent
- Ability to connect to a global database
With customer expectations and technology landscape transforming at a breakneck speed, organizations that are yet to hop on this change bandwagon will soon be forced to do so. The ones that fail to adapt will lose their competitive edge.