Can Banks Get Smarter?
Can Banks Get Smarter? Rethinking AI in financial services
By Rashed Haq and Conor Ogle, Sapient Consulting
Tackling costs has become a dominant theme for financial services organizations and a key consideration in any strategic decision. The cost per customer of operating a bank in today’s market has risen to the point that, all too often, it is prohibitive to delivering high levels of customer support.
A major contributing factor is the growing expectations from more demanding and savvy customers. They are aware of an expanding array of choices and highly value personalized guidance and services. Added to this, newer and more agile firms have disrupted the landscape, bringing these services to market quickly with the use of emerging technologies.
Maintaining the pace and keeping up with technology leadership has proved a challenge. For some firms, simply running the bank has required major ongoing investment and introduced challenging backlogs in technology and customer experience. The pressure of incremental regulatory compliance further compounded the issue, absorbing significant portions of budget and change capabilities. With this burden now easing, firms have freed up some budget and resources to invest in developing their propositions. For them, this is just the beginning.
Firms recognize the risk of being left behind if they tread water with the same approaches. Equally, they need to be cautious about how to initiate change so as not to waste resources and simply amplify existing problems. To thrive, they must align their efforts around a single common goal: achieve agile customer centricity at a sustainable cost.
Agile Customer Centricity and AI
The key is to strive toward creating a genuinely agile enterprise able to deliver the right support today and seamlessly accommodate future changes. Enter AI.
There is an enormous amount of activity in this area with Silicon Valley companies leading the way, investing billions of dollars exploring the many possibilities. Tech giants such as Google and Baidu spent an estimated $20 billion to $30 billion globally in 2016, according to research firm McKinsey Global Institute.1 However, financial services firms, historically ahead of the technology curve have been slower to implement AI in real practical terms. Ongoing high-profile investments reflect the industry’s enthusiasm for AI but many have struggled to effect real business change, which suggests it’s time to take stock.
There are many possibilities available to harness advanced data analytics, machine learning and intelligent process automation. Research firm Opimas predicts that AI will have a major effect on capital markets jobs, with over 20,000 more jobs in technology and data by 2025. As efficiencies grow, it predicts as many as 90,000 fewer jobs in asset management, 58,000 fewer in securities services and 45,000 in sales and trading.2 However, these numbers will pale in comparison to the opportunities to improve productivity and create true incremental client value.
It is easy for firms to pursue AI projects that simply garner attention without actually benefiting the business. For AI to have a meaningful impact in financial services, it must align with the core business purpose= and, specifically, deliver shareholder value. The customer is central to any firm’s success, so by focusing on customer centricity it can organize and prioritize its AI efforts to get the best results.
To achieve this requires agility. Firms must be innovators and think like start-ups, with a focus on speed, emerging technologies and data. The best AI strategies are those that enable firms to unlock their full potential and, ultimately, do more with less.
AI in Practice
In client-centric financial services organizations, AI can fundamentally improve business performance. The recipe for success is to enhance the advice they can provide at scale, reduce costs and then derive greater value from what they have. AI can help them to achieve this through three distinct yet interlinked applications: robo-advice, robo-ops and robo-alpha.
A growing challenge for firms is how to consistently offer high-quality advice at scale. That means providing personalized recommendations to clients that are relevant, suitable, timely and actionable. To do so requires them to truly understand each client’s needs and offer an experience that supports both digital and integrated human interactions.
AI can help by delivering advice directly to the customer or to the relationship manager, who then integrates it into their support. A key aspect of this is automating insight generation. With AI, firms are automatically extracting previously unknown insights from structured and unstructured data, based on predictive patterns and causal connections. This enables them to identify and act on recommended actions for improved competitive advantage. As such, AI is providing valuable resources as well as intelligence throughout the entire customer journey, from initial contact and onboarding through to ongoing service.
Morgan Stanley will be using AI to support its 16,000+ financial advisors by taking over routine tasks as well as augmenting advice.3 The project, “next best action,” will use machine learning to make recommendations and also track phone, email and web interactions to improve suggestions over time.
In the UK, HSBC is using robo-advice to deliver quality advice efficiently to customers with savings of less than £15,000.4 The UK’s high street banks previously withdrew from providing investment advice to this market following fines for mis-selling and more stringent regulations. Robo-advice offers an opportunity to close the advice gap at a lower-cost, conduct-controlled service.
Removing inefficiencies is perhaps one of the most valuable applications of AI. Poor data quality, evolving regulations and legacy IT structures all contribute to the complexity and time-consuming nature of many processes. As a result, firms are redesigning the workspace environment for colleagues and customers and automating the simplest tasks to minimize friction and operational risk.
Using AI to eradicate repetitive tasks liberates employees from the more monotonous duties that require little cognitive effort beyond common sense. Intelligent process automation (IPA) is AI enabled robotic process automation (RPA) and one such technology to reduce costs as well as headcount.
IPA essentially augments the workforce with a team of software robots. Common applications today include companies using chatbots to handle online client interactions and answer frequently asked questions, IT departments monitoring the health of network devices and operations and finance teams speeding up back-office work such as data entry and reconciliation.
In financial services, there are many further opportunities, particularly within data management. For example, an investment bank might have several manual processes to ensure any changes to instrument data arriving upstream are reflected in the downstream systems. IPA can transform this function, with robots retrieving data exceptions from the source system and entering the information downstream. This allows staff to reallocate their time to more valuable tasks such as better serving customers.
While robo-ops leads to some significant cost savings, other major benefits include quicker completion of processes such as customer on-boarding, improved accuracy throughout the customer experience and the ability to scale resources to meet changing demand.
IPA has been helping Deutsche Bank improve productivity and augment its workforce, according to reports. Applications in trade finance, cash operations, loan operations, tax and others have achieved between 30 and 70 percent automation, improving quality and decreasing risk. IPA has also reduced the time is takes to train employees.5 By encoding knowledge through robotics, that information becomes available to augment employees’ skills and guide them through their everyday responsibilities.
Meanwhile, Goldman Sachs has increased its use of complex algorithms with machine learning capabilities to streamline its trading operations. It now has two traders on its US cash equities trading desk, down from 600 in the year 2000. The bank applied automated programs, supported by computer engineers, in areas where it was easy to determine price and is extending them to the likes of currencies and credit.6
Having enhanced their advice and reduced inefficiencies, firms can focus on monetizing the value and knowledge held inside the enterprise.
AI can amplify the human intelligence and judgment within the organization by providing contextual knowledge and support. This helps both customers and employees perform tasks in increasingly simple and more effective ways. With more time, employees can mine data for greater insights and provide more value to their customers. Some examples of amplified human intelligence include expert assistants and intelligent searching.
One reported use case is JPMorgan’s AI program LOXM, which will execute trades across its global equities business.7 Unlike algorithmic trading, LOXM will learn from historic trades to set its own parameters, rather than rely purely on trader input. It will then execute orders at the fastest speed and best possible price. Its growing intelligence will mean it can tackle problems such as how to best offload big equity stakes without moving market prices. In the future, LOXM could get to know each individual customer so it can take into account his or her behavior when deciding how best to trade.
Building a Cognitive Bank
AI technologies cannot deliver the optimum results if they remain in isolation. Moving to a cognitive bank requires firms to enable robo-advice, robo-ops and robo-alpha across the entire enterprise. This means starting at the highest level, connecting all divisions, such as investment, retail, card and asset management services, with a uniform strategic approach.
Anybody building a new bank today would inevitably deploy this thinking from the outset, using the three applications to define the whole technological framework. However, firms can still apply this in retrospect to reconsider legacy setups and completely transform their business from the top down.
Determining the right combination of robo-advice, robo-ops and robo-alpha requires understanding the impact on humans throughout the entire process. It is not simply about catching up with competitors on the technology front. Instead, firms should weigh the investment required for each project against the effect it will have on both internal staff and external customers. By bringing it back to customer centricity, they can justify resources according to specific anticipated benefits.
Building this cognitive bank also means underpinning everything with a connected digital framework. Success relies on firms joining front- and back-office functions and data across the organization to deliver maximum efficiency and with the best possible customer experience. This will include, where relevant, adopting new technologies and processes such as data analytics, machine learning and cloud-based services.
The Future of AI?
With seemingly endless opportunities, AI has arrived at a crucial defining juncture. Countless resources have been invested and not all of them are delivering the results that financial services firms expect or need. To realize its full potential and deliver the best competitive advantage, firms need to rethink and refocus.
The future of AI—and its true benefit—lies in adding real, tangible value. That means focusing on what is most meaningful to the organization and harmonizing all AI efforts around that one single factor: the customer.
Robo-advice, robo-ops and robo-alpha encapsulate the many possible applications of AI, including easing onboarding, reducing inefficiencies and improving dialogues. However, in all cases, firms need to be asking if AI is helping them achieve more with less or if it is simply adding an unnecessary cost burden for little actual return. Capitalizing on AI shouldn’t require significantly increasing technology spend. It simply requires folding efforts into the ongoing business in a more meaningful way.
Future success will favor firms that focus on measurable progress and results. Whatever their budgets, they will inevitably win out – not just in the battleground of AI but also in the pursuit of the valuable customers of tomorrow.
Rashed Haq is the Global Lead for Artificial Intelligence & Data Engineering and a Vice President at Sapient Consulting. Over the last 20 years, Rashed has helped companies transform and create sustained competitive advantage through innovative applications of artificial intelligence, dynamic optimization, advanced analytics and data engineering.
Conor Ogle, Group Vice President, Financial Services, leading the Sapient Consulting practice for financial services in New York. Formerly COO and also CMO in universal banks, he advises incumbent and emerging firms to improve business performance and the experience for both colleagues and customers. His experience includes major programs of business transformation in highly regulated environments with a particular focus on the cultural impact of emerging technologies.
With Eugene Kanevsky, James Redbourn, and Joanna Wong, CLSA
AI/ML on the buy-side trading desk is a long-term program with short- and mid-term deliverables.
Aiden uses the computational power of deep reinforcement learning to improve trading results and insights.
BondDroid’s AI-generated prices are integrated directly into LTX’s pre-trade analytical tools.
Fintech firm has partnered with Jim Toffey, founder of Tradeweb Markets, to create LTX.