Fiscal and monetary support provided by the world’s governments over the past year have partially insulated people and businesses from the impacts of the COVID-19 pandemic. Lenders have also benefited, with the measures helping to limit loan defaults. Still, as the stimulus is switched off over the next six to 18 months, a higher proportion of customers will face difficulties paying for household expenses—as many as a third in several US states, while in some Latin American countries, more than half of families are already struggling to meet their needs.1 In light of these realities, several lenders are enhancing their customer-assistance and collections capabilities and, in the process, are boosting their resilience to adverse events.
By focusing on engaging with customers effectively, lenders also ensure that they are better aligned with regulatory requirements. As the effects of the pandemic play out, regulators are placing more emphasis on fair outcomes, checking that lenders do not “harass” customers through their collections processes. In all of their actions, lenders must be viewed as collaborative and transparent, offering customers every opportunity to find a solution. Measures such as debt moratoria, light-touch communications, and tailored messaging have become minimum requirements in many markets. Indeed, regulators such as the European Central Bank and the US Consumer Financial Protection Bureau have provided clear guidance on how institutions should pursue these kinds of goals.
In addition to external factors, there are strategic and operational reasons to modernize customer assistance, and these in turn may create significant bottom-line value. Customers are more comfortable than ever using digital channels, suggesting that lenders that provide smarter, more interactive, and more personalized services will perform better than their less responsive peers. Many customers will emerge from the pandemic stronger and will require assistance with borrowing and investing. Further, with margins under pressure, it makes sense to pursue the productivity and efficiency gains that come with digital-first solutions.
Leading institutions have shown that a digital-first collections strategy is a route to value creation. Some have seen reductions in nonperforming loans (NPLs) of 20 to 25 percent, alongside huge cost takeouts (and related increases in productivity), lower conduct risk, and more than 25 percent boosts to customer engagement (Exhibit 1). One lender shortened its average repayment time by as much as five times.
The customer-assistance capability should be built on a dedicated stack comprising digital-first engagement, advanced-analytics-driven decision making, and a scalable API-equipped technology platform. This digital-first approach, supported by a tribe-based operating model and agile development cycles, can form the basis of a more productive engagement, both online and in person. Of course, in the context of complex cross-border organizations, there are significant strategic, operational, and IT challenges associated with change. However, in terms of both customer engagement and economic performance, the benefits of effective implementation will significantly outweigh the costs.
Critical capabilities in the customer-assistance stack
In seeking to optimize customer assistance, it pays to focus on proven value drivers in three key areas: digital-first customer journeys, analytics-driven intelligence, and technology enablement. Each of these can play a role in reducing NPLs, optimizing and automating operations, minimizing risk, and fostering a positive customer experience. The trick is to get all three to operate seamlessly—combining elements of “art” and “science” to create a harmonious whole.
Unlocking digital-first customer journeys
Almost 10 percent (net) more people, and 25 percent more millennials, will engage with mobile banking post-COVID-19, a McKinsey survey of customer behavior during the pandemic shows.2 Similarly, fewer customers plan to use their phones to talk with staff.
With customers increasingly at ease with digital banking, it makes sense for lenders to make greater use of digital assets. However, they must go beyond targeted interventions, such as emails and texts, instead striving for a holistic and customer-centric collections capability with digital at its core. That means becoming “digital first” at every point in the customer journey, supported by contact centers and personal interactions.
A core ambition in rolling out digital frameworks should be to encourage higher levels of self-serve. Where permitted by the local jurisdiction, digital nudges through SMS or WhatsApp are less intrusive than voice calls and less costly. Digital-first solutions also enable tailored messaging for individuals and offer the chance to create feedback loops of continuous improvement on a microsegment level.
McKinsey research shows that digital-first customers contacted digitally make 12 percent more payments than those contacted via traditional channels. The resultant boost to consistency of communication and decision making will also create cost efficiencies and reduce conduct risk. Where customers are less digitally enabled, lenders should work to educate them about the benefits and migrate them to digital channels.
Lenders that have implemented digital-first solutions have seen multipercentage point upticks in resolution rates, threefold increases in monthly installment payments across portfolios, and higher levels of self-service. Their cost of collections has fallen by at least 15 percent. One bank improved its digital communications through advanced email campaigns. It made the emails more engaging and easier to navigate, and it included concrete options for recovery. This led to a more than 40 percent jump in email openings and a 150 percent rise in click rates. By redesigning its website and mobile banking app, as well as improving the collections journey, a Brazilian bank migrated 40 percent of its inbound clients from assisted channels to self-service. This led to both better engagement and lower costs.
Embracing advanced analytics and machine learning in decision making
A digital-first solution should be powered by an AI-led advanced-analytics stack. That will help the lender manage risk strategically across the key stages of customer support and optimize outcomes both for clients and the business. The foundational elements for implementation at scale include rapid model-development and recalibration cycles, open-architecture tools and techniques, excellent internal and external data streams, and the necessary talent to both develop applications and translate advanced-analytics applications to business needs.
The fuel for effective analytics is data, and lenders should strive to augment their models with new data sources and alternative indicators of risk. That means enhancing traditional metrics (such as loan characteristics) with inputs relating to customer behavior, spending patterns, and likely responses. Such data are more accessible in countries with open-banking data protocols. The aim should be to create a 360-degree view of the customer, and some leading banks now leverage more than 100 data sources, encompassing inputs such as local regulation, macroeconomic reports, and public-health records. In addition, they use data collected during the customer journey, which they combine with a high-frequency workbench to accelerate decision making. Insights can then be fed back into models via reinforcement learning, leading to more effective communications.
Based on data-driven decision making, advanced-analytics-driven intelligence can be injected into collections journeys to create a more responsive and personal service (Exhibit 2). Common models to support targeted interventions include self-cure, value at risk, collector pairing, sentiment analysis, choice of best channel, and best offer. In the context of a personal loan, the stack can be applied from predelinquency identification through to contact strategies in delinquency (including contact-intensity optimization), offer management, third-party arrangements, and recoveries and repossession.
Advanced-analytics capabilities are invaluable in helping lenders quantify risks earlier. Ahead of delinquency, for example, they can enable microsegmentation for value at risk, leading to the determination of appropriate treatment strategies. When customers are less likely to self-cure, lenders can offer persuasive redirection without the need for voice calls.
Analytics also can play a vital role in understanding customer needs and behaviors. Lenders should aim to contact customers first through their preferred channels (this alone can lead to a more than 10 percent increase in payments). Again, effective segmentation is key. However, traditional approaches, such as economic segmentation through months on book or outstanding loan values, are insufficient. Instead, machine-learning techniques can help lenders route customers to favored interactions. In this way, customers can be matched with the call-center agents who are best suited to their situations. In parallel, sharper routing and contact orchestration can ensure that customers engage consistently with channels that meet their needs.
Alongside the data science, there is a requirement for “art.” That means “thinking like a marketer” and showing that the customers’ interests are the same as your own. Use of plain, empathetic language can be a powerful lever, especially if tied to a real understanding of the customer’s situation and state of mind. Advanced analytics are vital sources of insight into these nuances, and they can help lenders respond quickly when Plan A doesn’t work. The current buzzword for customer interactions across industries is “personalization.” AI algorithms that learn customer preferences and how they evolve over time are essential to these personalized engagements at repayment moments of truth.
These strategies have been shown to pay off when implemented in practice. One leading North American lender adopted an analytics-driven approach to achieve an 8 percent reduction in charge-off losses. It focused on developing sensitive predelinquency contact strategies, for which it leveraged more specific and better-targeted treatments. The bank identified riskier situations based on value-at-risk modeling, backed by cross-functional teams and an agile development approach.
The Latin American operations of a global bank boosted cash collections by 20 percent after implementing an analytics-driven collections engine. The engine optimized multiple attributes, including value at risk, channel preference, contact intensity, and cost to serve. Through these attributes, the bank was able to create a customized approach on a segment-of-one basis, leading to increased take-up and better client retention.
In a dynamic-assistance strategy, there is no one-size-fits-all solution. Through deeper insights, lenders can move from approaches based on linear road maps to dynamic engagements that ebb and flow with customers’ needs, and which are self-reinforcing: better data breed effective engagement, which creates more touchpoints and new data opportunities.
Get the technology right
McKinsey research shows that about 30 percent of tasks in 60 percent of customer-assistance occupations are automatable using current technologies.3 This suggests that the conditions are already in place for the transition to a digital-first customer-centric operating model and key performance indicators (KPIs). From reporting, which can be augmented with self-serve and real-time applications, to robotics for frontline collections, there is significant potential to improve services, cut costs, and boost productivity.
Digital technology will assist customers both outbound and inbound and can act as a gatekeeper that coordinates omnichannel contact points. Where routine tasks are automated, time will be freed up for employees to engage in higher-value activities in the process of migrating from transactional to customer-centric roles. A necessary precondition, however, is to ensure that staff are trained to leverage analytics and still offer high levels of service.
A modern architecture would comprise an omnichannel offering of a full range of contact channels (including social media, text, mobile apps, and telephony, as permitted) and a digital platform that supports both device-independent and contact-center interactions, as well as the ability to integrate with data partners and service providers (for example, by offering chatbot solutions). Flexible-data and IT architecture is the underlying capability that can support orchestration, organization, and advanced analytics. In addition, an integration platform should bring together historic interaction histories and payments, the outputs of which can be fed into dynamic collections (Exhibit 3). The overarching aim should be to build a single IT framework for all tasks and processes, which can serve the customer from the moment a loan or credit is originated, through the predelinquency phase to debt resolution.
Getting there: Key enablers
The world’s largest companies are serial reinventors. The leaders of these businesses understand that a focus on collaboration (rather than documentation), team autonomy, self-governance, and rapid evolution are necessary conditions of ensuring that IT and digital capabilities are aligned with business needs. In this context, lenders should create a bespoke operating model that is equipped to work with agile techniques such as DevOps (bringing together operations and development engineers) and cloud-based solutions.
The model should be composed of cross-functional teams that are able to engage closely with the business to support rapid application development. There are various options on how to structure these arrangements. However, common elements include a collections tribe, made up of vertical squads (divided by channel, stage, and segment), and horizontal chapters. These might include collections specialists, channel specialists, and data scientists. The tribes should be supported by self-managing teams, which provide services such as reporting and control analysis. Outside of the tribe nexus will sit call-center management, channel management, and legal management, all overseen by a head of collections.
The new hybrid organization should be structured to enable an agile approach across a range of activities, from content development, including message tone and personalization, to customer interfaces, sequencing (frequency and order) choices, and channel strategies. Some lenders have shown that agile approaches to conceiving, testing, and scaling initiatives can accelerate the development process as much as fourfold (Exhibit 4). Two-week agile sprints (or four to six weeks, where necessary) enable teams to move quickly from pilots in areas, such as personalization, tone, and content, to rapid testing and implementation. The compressed time frame also helps them run more pilots in parallel, increasing “hit rates” in line with the intense cadence of activity. To translate these efforts into results on the ground, lenders need to get into a rhythm of rapid testing, continuous integration of learning, and incorporating results back into models.
These principles, when effectively implemented, can create significant value. One leading Eastern European bank focused on capability building, continuous development of minimum viable products, major IT renewal, and use of external NPL services. In one year, it achieved a 20 percent reduction in its cost of risk. A retail-focused bank in India leveraged agile approaches to implement a collections transformation across more than 15 levers. Over 12 months, it imported technology and design experts, data scientists, and general consultants to launch multiple at-scale proofs of concept. It added about 15 percent of customers cured through self-service and achieved a doubling of payments made digitally and a 15 percent reduction in cost to collect. One Latin American bank shifted toward an agile organizational model in customer assistance, building multiple squads focusing on optimizing recovery and engagement. This led to a threefold increase in A/B tests, enabling fine-tuning of the proposition and a doubling of take-up of new solutions.
When it comes to buy or build, there is no imperative to choose one or the other. Most leading players combine off-the-shelf components, from data storage to email platforms, with custom-built strategic engines and machine-learning models.
The transformation of the collections function into a full-fledged customer-assistance capability is a significant task, particularly at a time when budgets are under pressure. Lenders should expect a 12- to 24-month journey. However, without transformation, there is a high chance of collections underperformance over the critical next two to three years and for incremental disintermediation as fintechs target links in the value chain.
The good news is that it is possible to capture early wins, achieving progressive efficiencies that fund the initial investment, and eventually exceed it. Furthermore, there is no imperative for a definitive target state from the start. But with the world changing at pace, it makes sense to build the capabilities that enable agile adaptation over time.
Most institutions will not approach collections transformation from a standing start. Indeed, many have implemented data and analytics transformations as part of new customer-acquisition strategies. These strategies will provide valuable learnings, including informing the decision on whether to buy or build. Most importantly, however, decision makers must seize the moment and take bold action across the collections stack. With that in mind, it will be important to put a team of cross-functional practitioners in place.
For lenders that find the right combinations, the likely benefits will include optimized collections at a critical juncture in the economic cycle, more responsive customer services, longer-term customer loyalty, reduced compliance risk, and insurance against being left in the wake of more ambitious peers.