Back-office productivity requires more than Robotic Automation
With the ‘Go Digital’ trend taking over the banking and financial sector, there seemed to be inherent gaps in how the digital phenomenon is perceived and implemented in its totality. While it makes huge strides in customer interface, it still falls way behind in revamping the existing old-school operations and processes. My argument is that, if traditional banks purely pursue improvements in customer experience without fundamentally untangling their back-office spaghetti, they would end up similar to a lipstick on a pig.
A chanced upon an interesting opportunity to design a solution to help mitigate this back-office challenge that most of these institutions face. I am happy to share that, after nearly 18 months of continuous iterations, spending time with operations teams in banks understanding their challenges and receiving feedback on our approach and after few successful proof-of-concepts later, we are now ready with a platform that will truly help banks simplify their back office in rapid fashion, without slowing down their digital efforts to improve customer experience. In this journey, we also created a platform that is truly industry and business product agnostic.
Current State of Digitization in Banks
To understand the real challenges on the operations floor, we spent time with the back office staff of various banks in ASEAN, India and Middle East, actively monitoring their typical work patterns and learning about the applications and tools that they commonly use. Most of these banks had embarked on a transformational journey around five to eight years ago to digitize manual workflows using Business Process Management (BPM) products. The more tech-savvy ones also procured Document Management Systems (DMS) to store the documents in a central electronic repository to replace paper based document storage. While the BPM engines helped centralize operations by getting branch staff and channel applications to route electronic copies of transaction instructions to remote/offshore back office departments, the operations staff still ended up doing a lot of their work manually.
Large banks, leveraged the centralization that workflow engines brought, and moved work to low cost/offshore locations and declared their journey as a success. In reality though, it is a lot of Band-Aid fixes that is being continuously applied to keep their engines running. The information on transactions now gets scattered in data stores across multiple systems, partly in DMS, partly on paper, besides a whole set of email trails and excel sheets. Despite electronic workflows and electronic document storage, transaction inquiries are still cumbersome, as it is difficult to get complete information from one application. Investigations by auditors and compliance continue to take up significant time as they need to sieve through data stored in multiple applications.
IT departments continue to solve this problem by building more and more message interfaces between these systems, besides creating a master data warehouse where all transaction information is updated for central reporting. Without a well-defined and universal data standard across the different systems, bulk of IT enhancements results in Extract-Transform-Load (ETL) and data routing enhancements that take up a significant part of the Run The Bank (RTB) budgets.
The rush to be the leading digital bank has exposed the challenges in back-office IT. Valuable investments are taken up by non-productive RTB tasks that keep IT teams busy fixing the plumbing leaks, as the different systems undergo enhancements to meet market or regulatory needs.
Robotic Process Automation (RPA)
Today, Robotic Process Automation (RPA) is on top of the agenda of most senior operations leaders. It promises to simplify the back-office data input and minimize data integrity challenges by employing software robots that can be programmed to perform repeatable tasks automatically. More importantly, this can be accomplished without making changes to back office system. Leading vendors have either a tool that they offer or tie up with partners to offer this capability as a service. The classic use case, these vendors demonstrate to get the banks excited, is an existing challenge of updating a customer address in multiple systems. Even now, it takes anywhere between three to five days for certain large banks in Singapore to complete their customer’s address change formalities. Banks, sensing benefits, started to quickly pilot these tools in their operations floor to demonstrate quick productivity gains.
While we fully subscribe to RPA technology, we felt that banks are getting into a situation that is quite similar to their investments in BPM and Service Oriented Architecture (SOA) around eight to ten years ago. According to industry research, roughly around 20% of BPM projects have realized the benefits that were projected at the time of implementation. Given the organizational silos in banks, most ended up with multiple BPM tools with differing implementation approaches. BPM implementation in banks continue to take anywhere between two to five years with significant customization resulting in huge capex investments and long term maintenance overheads. A case in point is a bank in ASEAN, that now replicates the validations of the back-office system, in their BPM layer. Another large bank has abandoned its BPM version upgrade project as the heavy customization done on an earlier version, needs to be redone in the new version of BPM software.
We see this trend repeating again in the Robotic Process Automation tool implementation as well. Large banks are simultaneously experimenting with multiple tools across different departments and will eventually end up with a situation like the BPM implementation. Given the hype around this technology, IT vendors are commanding premium rates for what is otherwise a mundane programming task. Finally, the bottoms up approach taken for implementation to yield quick results, would eventually result in a long term robotic automation script maintenance overhead that is similar to the RTB costs, banks invest to keep interface plumbing in order.
In order to address the operational challenges, we wanted to get a complete 360˚ view of what goes into operational tasks and we were lucky to get the valuable time of a few supportive banks and staff as we went about the study.
While manual data input and duplication in data entry is one of the hurdles to automation, we quickly realized that it is not the only challenge. In fact, we met some experienced data input clerks, extremely fast in keying data with an eye for finer detail, quickly spotting anomalies in data provided by customers.
Despite the teams using BPM and DMS solutions, they end up printing scanned copies of documents as these printed copies are used to mark additional information on “how” a transaction is processed, which in most cases the IT systems are not able to support. The additional documents submitted by customer, the printed copies of email conversations of staff involved in the transaction process, printed copies of approvals from credit teams and other regulatory functions in bank are neatly organized in a folder and operations staff faithfully uses the IT systems as pure workflow and bookkeeping system. They find this to be the most efficient method as when queries come in at a later point in time – they have all information in one place to respond accurately. When we quizzed them on the existing search capabilities, there was a consistent query from most users – “Can we get a Google-like search” that searches comprehensively across all relevant systems and data.
There is another need to keep the paper trail going. Users find it convenient to sign and stamp on paper and record them in archives, rather than embrace digital signature and cryptographic validation tools.
Most alarming, is the lack of end to end visibility that senior leaders have on their teams as it is difficult to accurately measure overall operational efficiency, despite investing heavily into IT systems. A customer transaction constantly moves from back office system to another, making it difficult to measure the overall life cycle time. Consequently, unit pricing and SLAs are more of an approximation than based on actual data.
Finally, one important aspect kept lingering in our minds. The operations users were explaining their activities using application names, rather than the actual business process step that was required for the product. We realized that IT has successfully managed to get operations to adapt and use their systems rather than the other way around where the systems should have enabled the actual operational process, without users realizing which application or technology they were connected to.
The study gave us very good insights on the various challenges that need to be addressed in order to improve the overall efficiency of back office operations. While RPA is an important contributor to efficiency, we realized that the entire ecosystem needed to be aligned in order to realize the true benefits projected.