How Banks Can Use Advanced Analytics to Better Serve Commercial Clients

By: Stiene Riemer, Jörg Erlebach, Cesar Torres, Markus Wiemann, Alex Wetzler, Barric Reed and Ian Stokes-Rees

Banks were among the earliest users of information technology. They began by implementing electronic book-keeping of client accounts and have since used technology to collect a substantial amount of client data. With the advent of the ATM, they were among the first to introduce information technology into the retail environment. Thanks to information technology, banks can see which banking products their clients have, where their clients get their money, and how they spend it. Banks have yet to leverage the variety of data especially when working with SME and commercial customers. One reason they haven’t is that commercial banking is still considered a “people business” that traditionally relies on personal interaction between bank and client staff. Banks may also face technical challenges when implementing new technology, as they were among the early movers and are now being challenged to modernize.

BCG GAMMA designed its proprietary SmartBanking.AI solution with the specific goal of fully transforming the commercial banking operating model. Banks’ P&Ls have proven that significant profit uplift of up to 20–30% is possible when they are truly transforming both the client and the banker experience using analytics.

To make it easier for banks to benefit from data and analytics-driven solutions, we designed SmartBanking.AI (SBAI) to be deployable in two ways. It can be deployed as Analytics-as-a-Service (AaaS), so that banks won’t have to staff the teams needed to develop and maintain complex technology, and so that teams can focus on the digital strategy and transformation of the operating model. Or it can be deployed as a full Build-Operate-Transfer (BOT) in-house solution for banks that have the internal staff for maintaining and further developing analytics applications. Of course, it is also possible to start with the AaaS solution and switch to internal ownership once the technical requirements are met and staff with sufficient expertise is hired and onboarded.

Our solution is already successfully being leveraged by many leading banks across the world.

SmartBanking.AI’s Main Goal: Turbocharge Relationship Managers

To service most commercial clients, banks will appoint a relationship manager to serve as the first point of client contact. (Depending on the client’s financial needs, the banker may bring in additional product specialists.) Be an advisor — know the client’s business deeply and preempt their needs. To be successful, he or she must acquire a deep understanding of the client’s business model, its financial transactions, and its banking products and services. Gaining this understanding, however, can be complicated by several factors:

• The often-complex structures and connections between multiple legal entities within a single client relationship
• Available time for each individual client — Lots of clients need to be served
• The wide variety of banking products available for commercial clients
• Information that may be spread across multiple bank IT systems
• The lack of a consolidated view to identify specific patterns among vast sets of transactional data

To date, little has been done to create automation that can support the multiple tasks comprising most relationship managers’ jobs. In most banks, the manager must access multiple systems to get a comprehensive overview of a client’s banking products. Never an easy task, it is made more difficult by the disparate systems within the bank — and some of which may lack consistent mechanisms capable of mirroring the legal dependencies between client entities. When relationship managers perform specific tasks, such as identifying liability relations while setting up a loan contract, they often must work manually to assemble information that’s exclusive to this specific task, and which they will probably not be able to re-use in another context.

Empowering Relationship Managers, Not Making them Redundant

Relationship managers typically communicate intensely with their clients to gain a deeper understanding of their businesses and their financial needs. The complexity of the relationship manager’s job — coupled with the lack of perfect client data — has slowed the full automation of many of their tasks. But what may seem like a difficult challenge can also present new opportunities. Many relationship managers are understandably concerned that Analytics or AI might make their positions redundant. But rather than seeking to replace relationship managers, SBAI helps them work smarter.

By using statistical models developed to leverage a bank’s internal data, SBAI can, for example, identify sales opportunities, as well as clients that are at risk for attrition or credit default. By themselves, these models will be of limited use if relationship managers don’t understand why a specific model output is generated for a specific client. To make the results more transparent, SBAI displays the underlying data observations in the form of business insights that:

• Explain to the relationship manager why a specific client needs a specific product — or is at risk of leaving the bank
• Provide the manager with additional talking points for client conversations

In one concise view, the solution displays consolidated information on all the client’s products. This can save relationship managers time they might otherwise have to spend extracting the information from multiple systems — and give them more time to address client needs.

The SmartBanking.AI Solution Setup

Banks can subscribe to the SBAI solution as Analytics-as-a-Service (AaaS), where the bank provides monthly data extracts and BCG GAMMA provides development, maintenance, operation, validation, and recalibration. (The bank can access the results via a frontend application, which can be customized to interface with the client’s existing CRM solution.) Alternatively, the bank can opt to take over the models after development and from then on operate, maintain, validate, and recalibrate them internally. Or they can start with AaaS and then, at some point in the future, switch to internal ownership. With SBAI, clients are not “locked-in” to a single approach. Note that whether the client uses the solution as AaaS, deploys it on-site, or takes a more intermediate step, BCG GAMMA will always develop the initial models.

SBAI’s first layer combines information from all relevant bank systems to create a comprehensive data set of the bank’s commercial clients. This includes extraction of information and consolidation of the relevant entity levels that are accessible in the data.

To address relationship managers’ needs, SBAI presents three types of information extracted from the data:

1. A 360-degree overview of consolidated customer information: This is intended to provide an overview of the products the client currently uses — without the relationship manager having to gather this information manually from multiple systems. The overview also incorporates product specifics such as loan balances, as well as changes over time such as recent increases in credit-line utilization.
2. Leads for specific customers and use cases: These leads are created by different types of models that use the vast data pool created as part of the 360-degree overview. SBAI displays high-value leads only, i.e., prevents overwhelming the relationship manager by showing too many leads that only have a low likelihood of conversion. The leads may be customers who need specific banking products. Or they may be customers who, depending on the use case in question, are at significant risk for moving to a competitor bank or going into default on their loans. Clients are differentiated along two dimensions; if they show an unusually high likelihood of making a purchase or presenting a risk (low, medium, high, or very high), and by the dollar size of the corresponding opportunity or risk. This differentiation enables relationship managers to prioritize their actions to have the best impact on their customer base.
3. Human-readable insights from data: On their own, leads may be of limited use to the bankers. To increase their value, SBAI amends them by adding insights on customer behavior (e.g., decrease in balances, increases in credit line utilization, maturing competitor loans, etc.). SBAI will display only those clients whose behaviors the bank considers to be extraordinary, rather than simply “business as usual.”

Three Selected SmartBanking.AI Use Cases

In addition to providing relationship managers with an invaluable 360-degree client view, SBAI addresses these managers’ needs through the use of specific models and their corresponding insights. Although SBAI encompasses further use cases like pricing incl. Environmental Social Governance (ESG) adjustments, or credit adjudication, the following selected three use cases shall exemplify the concept of leads and insights in more depths.

Use Case #1: The “Next-Best” Solution

The goal of the next-best-solution use case is to identify clients in need of specific products, and then to issue these leads, including talking points and all the relevant supporting information, to the appropriate relationship manager. We train the underlying models using the bank’s historically observed sales events as well as other contextual information about the client. This information is derived using the bank’s internal client data. Explanatory variables are built from data that was available before the sales event. The resulting probabilities that a specific client would purchase a specific product are calibrated to likelihood buckets (low, medium, high, very high). Leads are displayed for those clients with a significantly increased likelihood of purchasing a product.

SBAI shows relationship managers the odds that the client would purchase a certain bank product, while adding logic to size each opportunity in terms of its dollar value enables the managers to focus activity on the most valuable leads. Again, SBAI shows the bankers only those insights that are 1) supported by the client’s data, 2) relevant for the specific product in question, and 3) extraordinary.

SBAI also creates leads using rules based on the bank client’s transactional data. When there is insufficient client data to train models for a specific product, rules may be the only lead source. Rules are also set up to suppress leads generated by the model or to suppress other rules, such as that “no loan should be offered to clients with a bad risk rating.”

The following graphic illustrates how such leads are presented to the relationship manager. The lead consists of the recommended product, the applicable client-specific insights that can explain why that client may need the product, and the opportunity size in dollars.

Use Case #2: Increasing Client Retention

The objective of this use case is to reduce the attrition rate within the bank’s portfolio. Profitable customers and those likely to ‘stick’ long term should be retained. With this awareness, the manager could then take whatever action he or she deems necessary when talking with the client.

For this particular use-case, the SBAI solution focuses on identifying as early as possible those clients that are likely to leave the bank. SBAI does so by using a machine learning model trained on previously observed attrition events as target, and on the client’s behavior data at least three to six months prior to departure as explanatory variables. SBAI issues alerts for those clients at the highest risk for attrition. The alerts are further enriched with insight messages that explain why the solution flags this specific client, and with the dollar value at risk if the client decides to leave.

Use Case #3: Automated Risk Monitoring

Credit risk ratings are one of the key metrics banks must determine to assess the probability that a client will default on a loan. The ratings are updated at least once per year for every lending client. For commercial customers, those ratings are typically based on financial-statement information supplemented by qualitative assessments. Sometimes they are also based on quantities derived from the client’s transactional data.

Typically, determining these ratings is essentially a manual process. Rating for larger loans must be signed off by more than one person and, therefore, cannot be fully automated. To determine ratings, banks usually perform multiple analyses, the number and intensity of which are not often differentiated by the risk posed by the client. If they are differentiated at all, it is primarily by client size.

The SBAI solution’s ability to calculate risk scores that are automatically updated each month does not replace ratings. The scores, however, can be leveraged to issue alerts whenever one of the bank’s client’s scores deteriorated significantly from one month to the next. An alert allows the relationship manager to investigate potential issues immediately without having to wait for the next scheduled (usually annual) review. Nor does the manager have to wait until more severe indicators become obvious to the bank.

Having automatic alerts in place enables the bank to differentiate existing scheduled review processes so it can select the intensity of analyses performed as part of its rating update process (depending on the current risk of each client). In many cases, there may be no need to scrutinize a client’s entire business model if their risk rating is low and essentially unchanged since the prior year. In effect, the alerts enable the bank and the relationship managers to focus any intensive manual rating determinations on only its riskiest clients.

With regard to the two previous use cases, risk alerts are accompanied by insights that indicate why the alert has been issued and what changes have caused the score to deteriorate. Thus warned, a relationship manager can investigate the client’s behavior.

Making Relationship Managers More Productive and More Successful

SBAI is designed to substantially simplify the way banking client information is extracted and, subsequently, used to provide relationship managers with valuable leads and alerts, accompanied by client insights. To ensure a successful design and increase solution uptake, BCG GAMMA has included relationship managers as part of the development process. Their involvement ensures that the solution supports the managers in their job, and that developers will receive continuous feedback as they refine both the solution and how it is best put to use.

We support the transformation of the operating model and the solution roll-out by redesigning the sales process and enabling the bank’s team to provide targeted and interactive trainings. If deployed as AaaS, all users receive ongoing technical support. In our experience, both relationship managers and product partners quickly see the value SBAI brings to their jobs. Every day, they see first-hand and up close how advanced analytics, transparently explained, and supported, increases their chances of success in the day-to-day job of managing their clients and expanding their bank’s business.

• “There is some really interesting information you can pull from transaction data. That’s something we wouldn’t have picked up without being able to leverage the data in such an accessible way. This is a great insight!”
• “SmartBanking.AI provides very useful insights… I have a strong relationship with my client: I talk to them often, and they take my advice… but I had no idea they would benefit from the banking products the solution suggested for them.”