The baby steps to achieve an automated universe in lending
Financial service providers, like lenders and mortgage banks, are facing serious competition from the new age fintech players and digital banks, which offer faster loan disbursements and superior customer experience. The ever-changing external factors, such as the pandemic, liquidity issues, benchmark rate fluctuations, evolving consumer expectations, among others, have additional influence on businesses and service levels of traditional mortgage banks.
A study by Ernst & Young states that 64% of consumers are turning to digital banks, or fintech, for one or more financial products or services and this adaptation rate is growing fast. This situation leads us to the question, as to how can traditional lenders not only survive and sustain but also maintain healthy margins when consumers are rapidly turning to digital lenders and fintech players? One of the levers that are most popular and widely utilized in this scenario is the deployment of automation, and in several cases, Intelligent automation.
Traditional automation and intelligent automation technology tools can effectively spruce up the services of legacy lenders, by digitizing paper-intensive and cumbersome origination and servicing processes, thus improving user experience manifold.
One important step when embarking on the automation journey first time is identifying the right processes to implement them.
The first few crucial steps to implement automation
The success of any digital transformation initiative is dependent on common business and digital strategies at the organization level and aligning all the initiatives/objectives to those strategies. The next step is to assess and identify processes that are ideal for automation implementation. The question that needs to be answered is - which function or process should be automated first. The automation journey is a long and evolving one, and sometimes it can be overwhelming.
To put things in perspective, the first set of processes identified for automation should possess the below characteristics:
- A prospective process is suitable for automation when the process is paper/document-intensive, rule-based, has repetitive process steps, and is data-driven
- The chosen process should bear a low risk to the overall business process. This is an important risk management step to ensure that in the case of failure of an automation project, there is minimal or low impact to the business, at least until a backup process kicks in
- The tasks being automated should have a medium-to-low dependency on the downstream processes and dependent functions
- Automation at its best must enhance compliance, or at least comply with minimum regulatory and legal requirements
- Automating a process must have a clear pathway to deliver business benefits, such as productivity increase, reduction in the cost of operation and the average cycle time, as well as increased accuracies, among other metrics
In mortgage origination, the post-closing process, which essentially is a QA/QC process to check the correctness and completeness of a closed and funded loan, checks all the boxes for automation implementation.
What do we really mean by automation?
Now that a process has been identified, the question is, which automation lever can be utilized to deliver business benefits. RPA-based automation can be engaged where the processes don’t demand intelligent automation solutions, like character recognition, analytics, and ML models. Intelligent automation solutions, on the other hand, primarily consist of AI/ML-based solutions, enabled with the process automation platforms like UIPATH, Blue Prism, Automation Anywhere, AssistEdge, and so on. Intelligent automation tools can help realize higher benefits from automation faster and can help sustain the value delivered through it.
The most prevalent intelligent automation solution today is a combination of AI/ML-based iOCR (intelligent Optical Character Reader) platform with RPA, which can help any document-intensive process deliver higher business value. Generally, an average loan package contains 150-200 pages to be indexed into logical document groups, where auto document classification can be leveraged. The data audit process, at a minimum of 100+ data points, needs to be audited for correctness. Key solution components like AI, ML, and RPA would help in achieving
- 30% to 45%* reduced cycle time, thereby increasing productivity
- 60%* increase in straight-through processing
- Helping the mortgage originator achieve capital replenishment targets quicker, like selling mortgages in the secondary market faster
*based on the internal studies and process observations
What is the way forward?
Once automation or an intelligent automation project is implemented, the focus must shift to sustain the benefits accrued. There is a checklist one can follow:
- Check the impact of the automated process on upstream and downstream processes
- Tweak the automation approach, if required
- Identify other use cases that fall in the same category, such as post-closing
- Replicate the automation journey with similar Intelligent automation levers
Similar automation strategy can be adopted for other processes within the mortgage origination and servicing functions, such as email inbox monitoring, closing fee audit, funding audit, expired documents audit, letter generation & audit or simple system report extraction and reconciliation processes can also be considered for the next round of automation.
These can be your baby steps towards a mature and much larger transformation objective. But let’s first learn to walk, before we run into the automation arena.