14
 min read

Rapid Development of Data Applications in Financial Services

Rapidly develop and deploy sophisticated data applications on modern data architectures with Low-Code/No-Code platform

Rapid Development of Data Applications in Financial Services

The Banking and Financial Services (BFS) industry generates vast amounts of customer transactional data and is also subject to the highest level of government regulations. Due to the sensitive nature of the data and regulatory requirements, the industry adopted the highest data security standards and the best storage technologies. However, when it comes to utilizing the data to develop customer insights and new product development, the industry has traditionally been slow to change. But the need for digital transformation, competition from Fintech startups, and new customer demands are changing that. As new financial models such as personalized finance, robo-investing, cryptocurrencies, blockchain, and usage-based insurance take root, the large financial institutions are waking up from their slumber to harness the power of their vast troves of data.

Knowing the customer, building customer loyalty, and developing products that the customer needs are now the most important success factors for any business, but more urgent for the BFS industry. These ideals can only be executed well with the technology and data infrastructure that supports a deep understanding of the customer combined with responsive communications, personalized experiences, seamless multi-channel interactions, and effective operational processes.

The Challenges of Data

Data technology has always been expensive to create and maintain. While the rapid evolution in new technologies has created much-needed capabilities, maintaining legacy infrastructure, the cost of upgrading and the fear of breakage has kept financial companies from leveraging the power of modern data architectures.

Now, modern AI/ML and Big Data technologies are making it easier to organize data, discover patterns, and derive intelligence that can revolutionize operations and sales. But harnessing these technologies can be complicated as rapid evolution and skills shortage can hamstring your organization’s ability to develop and maintain analytical and operational systems.

Enter Datamorph - Kwartile’s self-serve data transformation platform. Datamorph is a suite of tools that optimizes development of modern data architectures. It shields developers from the complexities of the implementation, reduces development time, and allows engineers to focus on uncovering intelligence within the data. The platform abstracts out the complexities of two critical areas – distributed computing and deployment environments – that represent the most complex and time-consuming tasks in the development pipeline.

Datamorph makes it vastly more convenient to develop and maintain powerful data and analytical applications such as Customer Identity and Profile Management systems, Omnichannel Behavior Analytics, Helpdesk Analytics, and Returns Analytics. Additionally, the platform is customizable and extensible, allowing engineers to implement any kind of complex business logic required for building custom business applications.

With Datamorph, a retail business can create streamlined applications to obtain deeper insights into customer behavior, create highly relevant product lines, optimize customer service, vastly improve customer experience, and achieve high-ROI on marketing campaigns– all with significantly reduced spending on IT skills and infrastructure.

Finance Industry Trends and Outlook

Unlike all other industries, the BFS industry did not see the Covid-19 related pandemic upheaval. And thus, the industry will not see a paradigm shift that is likely everywhere else. What will happen though is that the trends of the past decade – digitization, personalization and the shift from RoE focus to product innovation – will accelerate.

The BFS industry does have a lot of catching up to do, as it traditionally had been a laggard in the areas of multi-channel Customer Experience and Personalization.

To be clear, the industry is facing challenges on multiple fronts: Consider the following statistics and trends:

  • The US banking industry will have to provision for a total of US$318 billion (3.2% of loans)in net loan losses from 2020 to 2022. This is a relatively milder hit than in 2007.
  • According to an Insider Intelligence report, tech companies such as Apple, Amazon and Google could grab up to 40% market share of the $1.35 trillion in revenue fromU.S. banks. Similarly, Amazon’s Pay-in-Store could reduce interchange costs for merchants and drastically cut into the $90 billion annual revenue stream for payment card issuers and networks.
  • The pandemic accelerated digital adoption significantly. 44% of retail banking customers are using mobile banking apps more often. This pattern is also apparent in commercial banking – e.g., Bank of America’s business banking app experienced a 117% growth in mobile check deposits.
  • Digital transformation projects are now being fast paced, with implementation time being reduced by from years to just weeks in some cases. For example, banks implemented Paycheck Protection Program (PPP) software in mere weeks – something that has been unheard of in the past.

Data over the last 10 years clearly shows that BFS organizations that invested in digitizing their businesses over the last decade demonstrated higher agility and resilience in adapting toCOVID-19-led changes than others. This agility will be needed more and more in the near future, especially as Covid-19 uncovered shortcomings in many Banks’ technology capabilities.

However, legacy infrastructure and data fragmentation across the enterprise continues to weigh down banks’ digital transformation initiatives.

Capability Requirements

So, what should a financial services business do in order to outperform in the near future as?

Apart from making core product capabilities resilient, there are six areas in which BFS companies need to enhance their capabilities. These are:

  • Hyper-Personalization and Focus on User Experience – Generation Z and subsequent generations show little brand loyalty, readily switching to competitors if the products do not meet expectations. Financial service providers must build the capability to allow consumers to design their own suite of banking products. As digital wearables become ubiquitous, it is increasingly feasible for banks to offer targeted and on demand services such as payment, banking, and money transfer services to customers.
  • “Whole-of-Bank” or Cross-Channel Loyalty – As retail, investment and wealth management firms become increasingly integrated, it is important for the organization to optimize the total value of their customers, rather than focus on individual products such as credit cards or HELOCs. The shining beacons of this approach are tech companies such as Apple and Amazon.
  • Digital Transformation is the process of bringing all the organizations capabilities online and exposing them in a self-service manner directly to the consumer. Digitization enables operational efficiencies, enhance speed-to-market and deliver superior customer experiences. To be sure, BFS organizations are well down the Digital Transformation journeys and are cutting down spending on branches to invest in self service digital channels. This needs to continue at a sustained pace, especially in the area of developing data capabilities.
  • Adopting to new monetary concepts – Cryptocurrencies are challenging the fundamental economic value provided by banks. Blockchain, which originally enabled crypto, is now revolutionizing a vast array of financial operational systems and business models such as peer-to-peer lending. Blockchain based registries are eliminating intermediaries and introducing significant efficiencies across all systems-of-record. As cryptocurrencies threaten the role of fiat currencies and blockchain’s incursions into core bank functions, BFS organizations need to rethink their fundamental business models for the long term.
  • Collaboration with Fintechs – Once seen as disruptive competitors, Fintechs are proving to be the friend of incumbent financial firms. This is because of the symbiotic relationship between the two – big banks need the technology agility while Fintechs need the access to markets.
  • Adoption of Robotics and AI/ML – As the deluge of data increases complexities and competitive forces require improved operational efficiencies, BFS firms must increasingly adopt Robotic Process Automation (RPA), and AI/ML technologies. Critical back-office functions such as Fraud monitoring, compliance, and customer service are uniquely amenable to RPA and AI/ML.

From a systems standpoint, all of the above capabilities require mature data integration and management practices supported by the latest data processing software and infrastructure. These include:

  • Cloud Data warehouses
  • Next-Gen Data Lakes/Delta Lakes
  • ELT Systems
  • Dataflow Automation
  • Self-Serve Analytical Systems
  • AI/ML Frameworks

More specifically, retailers need to embrace the paradigm shifts happening in data processing technologies, as illustrated below:

System Complexity

The above defined technology stack has great data manipulation and processing capabilities, but requires a highly skilled data team to organize the data, create processing workflows, and build operational data systems.

Add to this, advances in AI/ML systems which require complex training systems and a multitude of data stores for scenarios, and IT development operations can become unwieldy.

It is well known that about 85% of a data scientist’s time is spent on just identifying data sources and pre-processing steps. Similarly, application developers spend much of their time in dealing with the complexities of distributed computing environments (i.e., the multi-cloud),and the complexities of deployment environments.

The Solution

Kwartile’s DataMorph platform is a Low/No-Code Data transformation platform that reduces the hurdles to developing and maintaining modern data applications. The platform offers a powerful set of tools to create and maintain a modern data processing framework (see Figure 2):

Challenges of Modern Data Architecture

While the Modern Data Architecture solves many problems of data organization, data processing, data accessibility, and data governance, it creates a new set of challenges for developing data applications, including the following:

  • The complexities of
    - managing cloud-based data warehouses
    - maintaining deployment environments
    - implementing an optimized DevOps framework
  • Additional complexities of ingesting and transforming both real-time and batch data simultaneously
  • Inflexibility of Visual Tool Environments
  • Non-standard coding methodologies (SQL mixed with polyglot coding)
  • Non-standard logging environments

Due to all these challenges, the data applications programmer is faced with a steep learning curve. The underlying technologies are also advancing rapidly, making it even more challenging for developers to keep up.

What is needed is a way of reducing the quantity of coding required for developing and maintaining data applications, and managing DevOps, cloud deployments, and security frameworks.

The Datamorph Platform

Datamorph is a low/no-code platform that is optimized for developing data applications. By using the platform, developers can improve their productivity by 50-90% while reducing operational costs between 20-60%. In the long term, Datamorph provides many other benefits for software maintenance including automated version control, dependency management, and application security management. This leads to high code maintainability and improves the long-term economics of application development even further for companies.

At a high-level, the specific features and benefits of the Datamorph platform include the following:

  • Abstracts out the complexities of distributed computing
  • Abstracts out the complexities of deployment environments
  • Provides Visual tools for no code development but also supports custom code development
  • Fully extensible and flexible – Developers can add their own custom templates and code libraries
  • Includes deployment components for all leading cloud environments as well as on-prem environments
  • Sophisticated management of Type 2 Slowly Changing Dimensions into the warehouse
  • Data Skew analysis, which often becomes the nemesis of distributed systems
  • Built-in support for regulatory data compliance requirements including GDPR and CCPA deletes

Also, the platform provides many productivity enhancing tools including the following:

  • Design templates and architectural components for common use cases
  • Fast deduplication of data in streaming and batch mode
  • Built-in monitoring tools with unified interfaces
  • Rich data-quality detection library based on configurations
  • Provides pre-defined components and the ability to create customized component libraries

Case Studies

While the Datamorph platform is appropriate for almost any possible data application development scenario, some more representative Financial Services use cases include the following:

  • Real-time Stock Market Insights
  • Portfolio Risk Analysis
  • Customer Profile Management
  • Customer Behavior Analytics
  • Hyper-Personalization
  • Fraud Detection and Prevention
  • Help Desk Performance Analytics

Below are three representative case studies that the Datamorph platform has helped implement at BFS customer sites.

Customer Credit Information and Profile Management

It is no understatement that consumer credit information is the lynchpin that powers a vast array of consumer-focused financial products. From home loans to retail purchases, the ubiquitous credit score powers the risk assessment models that make modern finance possible. The Customer Id and long-term profile are critical information repositories required for making credit information products possible. However, the lack of a single consistent view of the customer, as well as alternate views of different segments, limits the market for consumer financial products. As an example, traditional credit scores are not applicable to the underbanked segment such as immigrants, even though their credit-worthiness is quite high. This can be ascertained by evaluating their online interactions and behavior.

Our customer - one of the world’s largest credit reporting and marketing agencies -faced significant challenges in unifying the various sources of data and create usable versions of the truth. Their IT team spends an enormous amount of time and effort in data wrangling and stitching together multiple internal and external systems in an effort to obtain a single, “best version of the truth”, and even then, only with poor results.

The emerging category of Customer Profile Management applications use AI/ML techniques to determine the identity of the customer without having to build traditional and difficult-to-maintain data models. Kwartile’s Datamorph platform enabled our customer to utilize AI/ML libraries and build new data applications significantly faster, with higher quality and reduced coding.

Fraud Detection and Prevention

Electronic fraud includes several types of transactional fraud that affects BFS companies. These include credit card fraud, billing fraud, and stock transaction fraud. Managing fraud to low levels not only creates financial benefits, but also improves customer loyalty. Creating an effective fraud monitoring and mitigation process requires processing large amounts of static and dynamic data to uncover patterns. Traditionally, these patterns are first surmised by human experts and then incorporated into the detection algorithms. But the effectiveness of such methods is lower as fraudsters keep changing tactics. Now, with the availability of sophisticated machine learning algorithms and AI, much of the process can be automated.

The Datamorph platform enables BFS organizations to create sophisticated fraud detection and prevention applications with built-in AI/ML and predictive models. The platform also comes with built-in tools to develop analytics applications and operational systems for managing, analyzing, and improving the returns process.

Customer Behavior Analytics

The core marketing functions for a retail business are improving customer conversion rates, reducing churn, and personalizing campaigns to increase revenue. A good customer behavior analytics system is a required to measure and improve the effectiveness of the core marketing programs.

The Datamorph platform enables integration of multiple data sources, near real-time processing of customer interaction data, and development of predictive analytics and recommendation engines, in a fraction of the time required with conventional software development methodologies. Built-in templates and algorithm libraries enable significantly faster implementations.

Risk Analysis

Risk analysis is core to every financial services business from insurance to consumer credit and retail purchases. Analyzing the risk of a transaction requires a large number of input parameters. Such input parameters include native transactional information such as the size of the order, and the number of currently active trades, and non-transactional data such as social media sentiment or monetary policy news. But as the number of input parameters increase, the complexity of the underlying algorithms increases exponentially, and in turn making the computational hardware expensive or impossible to solve the problem. Data scientists therefore keep the number of parameters to a low number while trying to get the best solution under the constraints.

With the great advances in AI/ML algorithms and hardware chips it is now possible to analyze transactions with even millions of parameters. Cloud-based data processing hardware also reduces the need to build expensive infrastructure. The Datamorph platform enables the development of sophisticated Predictive Modeling applications with built-in or custom AI/ML algorithms and the deployment of cloud-based hardware while abstracting the complexities of both the software development as well as the deployment parameters. This greatly reduces time-to-market and frees up developers to focus on developing intelligent solutions.

Benefits Achieved

The rollout of the projects provided multiple benefits and capabilities to product marketing and sales teams, as well as to the operational organizations including IT and Customer Service.

  • For one of our customers, the cost savings from just the initial implementations alone, ranged from $30 M to $50 M for each new credit bureau that is launched. Additional operational savings of about 20% per year were enabled by the better development framework.
  • For most of our customers, the time-to market for new product launches is typically reduced by 30% to 50%.
  • On the Data Science front, developing a predictive analytics solution requires only 40% of the time previously required. With the democratization of data that is enabled by the platform, data scientists obtain easy access to customer or transaction profile data in near real-time and can perform analysis to determine what aspects of deeper inquiry to conduct next.
  • Amongst BFS customers that already have mature data management practices, when Datamorph is added to the mixture, IT spends 85% less time in data unification, and 50%less time in developing and deploying new applications.

Support and Services

Kwartile offers a robust set of solution and support services for the Datamorph platform. Our Product Support services offer multiple plans for Enterprises, Software Development professionals, and Partners. The support services include technical help during the warranty period, bug fixes, and regular version upgrades.

In addition, our professional services team offers data strategy, design, and application development services. Kwartile Consultants and Data Engineers possess a combination of deep and practical technology expertise, industry expertise, and strategic product management experience