KYC – How well do I understand my customer’s experience and their data to Automate!!

                                    The Data and its Intricacies

Customers are the primary force that drives any industry ahead. ’Know your Customer’ (KYC) is an integral part of a customer's journey. KYC needs to be done while onboarding a customer. In addition, Re-KYC can be triggered for certain customers based on factors such as audits, risks etc. To ensure better customer experience a simple and smooth processing of KYC documents is required. Automation of KYC process is seemingly increasing with the main objective of ensuring faster and accurate digitization, effective verification, and shorter processing time. It is imperative for organizations to find ways to reduce the time and  spend on KYC as the average KYC process takes more than a month to complete and costs companies billions of dollars.

Complexities in KYC Process

Based on my experiences in the KYC space, I gathered that the KYC process is labor and time intensive as well as complex. Back-office teams do not just validate and verify the documents, but also spend effort to source the required documents and then identify and extract the correct content to ensure compliance. When the documents are of not good quality, the team needs to reach out to customers to reupload the documents to trigger the process again. In this fast-paced world, almost all data is present digitally and hence the call to leverage AI and Automation to make the process efficient.

 Why does it take so much time, effort and money for the KYC process?

1. Presence of 3 V’s in KYC data

● Volume owing to–

• KYC process is not only for a new customer but could also be for existing active customers. 
• Organization’s requirement to standardize the KYC process for all historical active customers and their data to enable ease of future re-validation.
• Large document running to hundreds of pages and information spread across these.

● Variety in –

• Document Types (photo IDs, Documentary and non-documentary evidence etc.)
• Document File formats (pdf, images, excel, word, emails etc.)
• Document content (structured - E.g. Passport, Driver license, Country specific Identity cards etc., unstructured – E.g. Registry documents, Company specific documents (Client letters, Company Ownership evidences) etc., semi-structured - E.g. Beneficial documents from 3rd parties, Financial statements etc.)

● Veracity because of–

• The spread of similar information across and within the documents.
• Need to Identify and extract the relevant objects of interest based on the business use case across this similar spread of information.

 2. Identification and Classification of document types and Extraction of document content

• The variety in document types and overlapping features within and across documents induces complexity in identifying the type of document (classification). E.g. Photo Identity cards, such as Passport, Driving License, Country Resident Card etc., vary in structure but are similar in content to be classified. Post correct identification of type of document, the relevant content has to be extracted.
• Presence of unstructured content with region (country), bank and customer-specific variations in documents. These variations diminish the generalization approach relating to retrieving the object of interest from the documents (extraction). The different entities in an organization or the senior management details could be represented differently in the documents and in different places in the same document (Ownership evidence, Financial statements , Registry documents). E.g. Ownership evidence of companies are present in different forms, like - flowchart depicting organizational hierarchy or a table detailing the ownership evidence or a free text in a document with needed information for parsing and retrieval.

  3. Data Privacy, Security and Region based document processing 

• Everyone knows that KYC contains PII and sensitive data. Therefore, any cross-location transfer of data for processing or consumption will need to adhere to the law of land with data security as prime focus.
• Many banks maintain data centers across each region and sometimes specific to a country as well in adherence to the PII and security restrictions and agreements.
• The KYC documents are sourced across regions and countries with customers across the globe, the documents are multilingual and the KYC automation solution needs to support multilingual processing for identification and extraction of needed content.

 4. Business Rules and format standardization

• A number of business rules need to be applied on extracted content from the documents which denote the industry and bank-specific standards.
• The format of representation of extracted data needs to be in a generalized form so as to be easily consumed by other 3rd party systems.

To handle the complexities and provide better user experience with value to business, any solution need to

1. Validate the quality of user uploads immediately and verify the authenticity of uploads(as applicable and possible).
2. Source documents / information as much as possible based on earlier data provided or from other third-party repositories (e.g. Swift KYC, Banker’s Almanac, Government Registries or Commercial Registries) who are authorized to share the data for risk, KYC and re-KYC. There will be a considerable reduction in wait time / manual requests for documents to be uploaded.
3. Automate the entire document classification and extraction (Potential increase in efficiency by 15-20% of overall validation time ) with human in loop (HIL) ability for reviews / approvals.

With all the above groundwork, how and where does automation aid this KYC journey? Is it enough to just have a process workflow and silo AI models to help identify the type of document and get the desired content? Based on my experiences in this space, we need something beyond AI standalone models and direct process workflows. This is where Hyperautomation fits in with its role of laying out an end-to-end solution encompassing all dimensions as needed for this industry use case.

More insights and perspective on KYC and Hyperautomation to follow soon!!! [TBC]