Where can automation be introduced into KYC?

There are a number of areas in which automation can play a role:

• Collection and ingestion of data and documentation from either public or paid sources
• The identification and verification processes using OCR, biometrics, machine learning and API
• Risk-based decisioning  of clients based on risk models and rules engines
• Workflows based on specific rules and scenarios, e.g., escalations and triggers
• Identification of document types and data extraction using machine learning (ML) and natural language processing (NLP)
• Reduction of false positives from screening hits (Sanctions, Adverse Media, or PEPs) using ML and NLP to contextualise and eliminate clear false positives, duplicates, previously cleared results, or not relevant results.

There are of course, some challenges:

• Gaining agreement and buy-in to automate KYC This is clearly a critical first step. Organisations have lots of other competing priorities. Do they consider KYC automation to be critical? Do they consider other initiatives to be more important? Can they secure the budget and sponsorship of the senior management team to undertake this transformation?
• Understanding what is actually required Many organisations like the idea of KYC automation, but do they understand what they are trying to achieve with automation? And do they understand what the benefit will be? Do they know which processes, policies, teams and technology are impacted? Do they have clearly documented and well understood policies, processes and procedures?
• Agreeing an overall approach There are many different ways to tackle automation and this can be a major challenge to progress if the approach is not clear and agreed. For example, have they got an agreed scope, plan and method for the programme? Are they clear on whether they will use 3^rd party vendors, and if so, for which parts? Do they understand any critical dependencies or constraints to deliver automation successfully.
• Being able to deliver effectively If an organisation is committed to KYC automation, do they have the right organisational sponsorship and teams in place and engaged in order to successfully deliver? Do they have the right skills and experience in the project team? Are all stakeholders bought into the process?

In our experience, there are 7 principles to follow to ensure your KYC automation programme is a success:

1. Think transformation, not automation – use the opportunity to transform and improve the way you do things today, rather than just automating existing processes. Explore the opportunities for simplification in order to maximise the benefits of automation. Measure the current KYC processes against a set of KPIs and demonstrate the benefits as you progress through your transformation programme.
2. Understand your requirements first – you need to ensure you know what you’re trying to automate before you start. This sounds obvious but it is not always the case. In the first instance, this requires clearly documented and well-structured policies, processes, and procedures. You will also need to understand the data and documents that you collect and associated metadata. You should then set a clear scope and set of objectives for your programme so that everyone understands what you’re trying to achieve.
3. Start small and deliver quick wins – be ambitious with your automation programme but make sure you don’t overreach too early. Try to break this programme of work into smaller phases. Focus on specific areas of automation and build a roadmap that provides a clear structure and plan of attack. Try to achieve some quick wins and show the benefit of automation in order to show results and gain buy-in from the organisation.
4. Use a cross-functional team – at a minimum, you need representatives from across operations, compliance and technology (likely front-office too). Engage with the users on the ground who will be impacted to fully understand the requirements.
5. Don’t underestimate the data challenge – automating any KYC process will inevitably require a huge focus on data. Incorporating new data or taxonomies into an existing data model and existing systems can be the trickiest part of this puzzle, so make sure you focus on the data aspect from day 1.
6. Agree third party approach – some organisations look to automate processes using only in-house resources, while others seek out 3^rd party vendors to solve their automation challenge. The balance is always somewhere in between. Identify areas where a 3^rd party can deliver automation in a more cost effective or higher quality way, but be targeted and specific about what you want to achieve. 3^rd parties can be used particularly effectively where this is a discreet process that can be segmented and there is a simple way to interface with that solution e.g., via API. Ensure you identify early where you believe 3^rd parties will be required to support and which parts you will tackle in-house.
7. Test, and test again – this is clear with any technology-based initiative, but it is especially important when processes are being automated. Make sure that every scenario including edge cases are considered and tested. If they are not picked up during testing, it could potentially be a long time before anyone realises something is wrong. Engaging a wide user group for testing is critical.