To achieve real results that materially exceed the cost of implementing a big data platform, banks need to set a specific agenda for developing high-value applications.
As information-based competition explodes in the digital economy, banking executives increasingly are asking what kinds of investments they should make in “big data” — a complex marriage of vast customer information, technology and analytical methods. The expectation is that over time in the all-digital world, this powerful combination will provide essential fuel in optimizing bank customer relationships.
But big data also poses a huge stumbling block for those banks that get lost in the details. The technology team can get too far out in front in considering a new platform before the line of business team (e.g. retail banking) has developed a strong sense of the high-value business applications that should be pursued. Plunging ahead, the bank then gets locked in a chase to justify new systems — a solution looking for a problem to solve — as happens too often with major technology and process trends (e.g. CRM in the ’80s and ’90s).
To achieve real results that materially exceed the cost of implementing a big data platform, banks need to set a specific agenda for developing high-value applications. For smart financial services players, that agenda relies on mastering the art and science of managing the “customer journey” — analyzing, predicting and then matching the progression of customer needs relative to the progression of the customer relationship with the bank.
Keys to a Smart, Fast Start
There are many pitfalls to be avoided with big data. Based on Novantas research and client work, there are three management keys to a smart, fast start:
1) Approach customer journey analytics and big data as an experiment. Despite the allure of big data technology, adoption is still in its infancy in financial services. A recent Novantas survey of 10 top institutions identified only two that are seriously engaged in converting key functions (core decision support or operational data stores and analytics) into a big data environment.
Management should be cautious about approving a “big bang” investment without first thinking through potential business use cases and the trained analytical resources needed to operate in the big data world. Significant advance work is needed on use cases, not only specific applications of big data but also the significant operational changes it imposes on traditional analytical environments. Applications related to the customer journey, especially relationship deposit pricing and managing the digital banking relationship, are a great place to start.
2) Leverage packaged technology approaches (outsourced or in-house). Third-party providers have totally outpaced in-house developers. Players like Amazon offer easy-in, easy-out programs with all the technology capability needed to get the bank up and running quickly, without the headache of infrastructure management. An elastic cloud platform for initial experimentation also helps IT teams to ramp up core skills for big data platform management. A small, packaged in-house solution is also available, but again, it should be appropriately sized for experimenting.
In many cases, an initial investment of roughly $500,000 can secure an environment that can handle all of the customer analytics for a $50+ billion institution, including 10 years of customer transaction data, product and channel behaviors, plus outside customer wallet information. A small or large bank can contract with Amazon and operate such an environment for some $30,000 a month, with no long term contract.
3) A governance model is critical. Even in the early experimental stages with big data, project governance is critical to ensure the ultimate adoption and sustainment of any program. One example is with derived metric definitions (a form of business metadata).
One of the promises of a big data environment is speed and access, with a lot less time spent organizing data into traditional relationship tables. In theory, an analyst can easily access and develop metrics, features or models, using new, analyst-friendly programming languages.
The downside risk is that uncoordinated efforts can compromise the consistency of derived metrics, models and analytics. Without clear definitions, say, for net present value, there is a risk that various analysts (called data scientists in this new world) will rely on differing calculations, creating myriad inconsistencies.
For one model to be used in a big data application related to the customer journey, for example, Novantas developed more than 500 standard metrics. Such definitions and controls of business metadata enable users to speak the same language, customize definitions to their liking, maintain version control, and always be able to trace back the sources and uses of data, metrics and models in their analyses.
This is not trivial. Business metadata governance standards and the right tooling need to be in place from the outset, otherwise applications will never be able to move from the experimental stage into production.
—Darryl Demos and Kaushik Deka
The goal is to determine analytically what product/pricing combinations to offer to which customers, through which channels and at what time, so as to optimize both the value provided to the customer and the value of the relationship to the bank. This ability is increasingly important in a digitizing banking market that is characterized by dwindling face time and growing customer reliance on impersonal electronic channels.
Most immediately, there are a few specific high-value applications where big data, intelligently implemented, can provide banks with significant returns on investment, including: 1) deposit relationship pricing; and 2) multi-channel customer relationship management, the latter including marketing, targeting and offer management.
Such applications capitalize on the special abilities of big data, but also pose special requirements. In the Novantas experience, two warrant special mention:
Deep business context matters. Banking expertise is critical in posing informed hypotheses, interpreting preliminary findings and refining further exploration. Many big data initiatives lapse into an admiration of technology. But systems cannot be expected to identify opportunities strictly on their own — expert managerial guidance is needed to properly point the technology at big industry challenges.
Execution and change management matter. A big data analytical program will fail in isolation. Operational integration is required from the outset to ensure that execution realities are considered as opportunities are explored. To successfully make the leap from “experimentation to implementation,” the management team needs a roadmap that anticipates and addresses the changes that a big data transformation will impose, including people, technology and governance.
Rationale for Customer Journey Analytics
Retail banks have been operating in a weak recovery era following the recession. General market growth has been modest at best, and the emphasis is on expanding and retaining current customer relationships. Undercutting progress against this goal, however, is the erosion of branch face time as customers relocate their banking activities online. Novantas research has shown that banks are far less effective at developing relationships and selling products to customers who seldom use the branch.
On the bright side, digital banking offers extensive possibilities for personalized offers that can be pushed to the customer via multiple channels at low cost. As well, all of the data from product and channel usage (purchase and payment transactions, payments media, the flow of funds among accounts, mobile and online activity, etc.) can be captured and analyzed to provide insight on customer needs and sensitivities.
But to make sense of the information and leverage its possibilities, it must be organized in a way that reveals salient patterns of customer behavior over time. This includes looking across channels, transactions, products and promotions.
In this context the use of big data is twofold: 1) the incorporation of data from outside the bank (third party, web-derived) plays an increasingly important role in augmenting internally-generated knowledge of the customer journey; and 2) a big data platform delivers the processing power needed when combining years of in-house data with external data, plus the quick-turnaround analytical insights needed to guide decisions.
Major banks have a growing list of priorities that require advanced technologic/analytic capabilities with customer information. Two immediate applications are customer pricing and management of the digital banking relationship.
Customer pricing. Courtesy of the web, customers have a lot more information about deposit and credit offers, plus more reference points on competitive prices with online banks, eroding geographic fencing and reducing margins. Meanwhile digital delivery is exploding the number of offers customers see, requiring more valuable (read expensive) offers to grab customer attention for deposit and loan growth. Making these offers broadly available is uneconomic and often results in negative marginal contribution (incremental loans and deposits actually reducing bank profitability).
Big data provides a powerful lever in dealing with this situation, enabling banks to develop models that can more granularly target customers based on their responsiveness to price and the likely retained value from that offer. This capability delivers new opportunities to the bank:
- A new level of precision targeted pricing, to reach those customers for whom price will drive incremental volume, and for whom the persistence of the deposit and/or utilization of the credit will yield positive economics over the estimated lifetime of the balance (including the incremental cost of the incentive).
- New opportunities for customer acquisition, using digitally-refined targeting to find prospects with similar profiles — reducing the reliance on broad campaigns that erode value by paying incentives where they are either not required or where the customer’s behavior will likely not deliver the intended economics (e.g. promotion hoppers).
Having identified target populations based on the appropriate offer (price or otherwise) for balance consolidation, big data’s machine learning capabilities equip the bank to rapidly test multiple dimensions of offers (price level, channel, message, conditions) against target populations, not only to optimize who should get an offer, but also to pinpoint which type of offer is the most economic over time.
In select practical cases, big data has already helped banks to target for both promotional prices and product offers, and, in some cases more importantly, who not to target for promotions. The full slate of loyalty and propensity analytics, combined with past behaviors, helped these progressive players to avoid “waking the sleeping giant” of easy-to-activate, rate-sensitive customers. To fully realize the potential of these analytics, the final enablement is developing the direct communication strategies required to elevate response rates to levels as high as mass marketing from relevant segments.
Managing the digital banking relationship. The management of digital banking has evolved in loosely-coordinated stages, with plenty of internal overlaps, conflicts and gaps. It has been difficult to coordinate marketing, sales, multiple product teams and multiple channel teams for a cohesive outreach to multi-channel customers.
Mastery of big data is essential in helping banks to reach the point where a single data management platform carries the load in: 1) compiling internal, vendor and web-derived customer data; 2) applying a consistent set of rules and models: and 3) supplying the continuously-updated output as a basis for coordinated outreach.
With the sales funnel, for example, such output can inform the use of a demand side platform, a system that allows the buyer of digital advertising inventory to manage multiple ad exchanges and data exchange accounts though one interface. It also permits a tight integration of campaigns and cross-sell prompts across bank channels (website, mobile, e-mail, branch, call center, ATM, direct mail).
Complex Management Challenge
The backward approach to big data is to start by evaluating systems and features, with the expectation of finding economically viable applications over time — the proverbial “hammer searching for a nail.” Industry veterans are still smarting at the remembrance of how this approach led them into a swamp with CRM, which required a decade of trench work to fulfill the promise of delivering all relevant customer information to one place.
This time around the bar is even higher, given big data’s promise to not only deliver all customer information to one place, but to analyze all customer information from all over the place to deliver actionable insights. To do so, big data depends on an entire ecosystem of data, technology and analytics, which must evolve together.
The situation presents a complex management challenge:
- Developing tangible, high-value business applications through a customer journey analytics framework;
- Doing so in a way that leverages the processing power and cost dynamics of new technology; while
- Adhering to sound governance practices so that results are scientifically transparent, replicable and auditable.
Though difficult, these steps are essential in preparing the bank for a changing market. As more customer shopping activity and transaction volume diverts from the traditional branch to online and mobile channels, banks increasingly must develop effective personalized applications that will work in impersonal circumstances. The challenge is compounded by new competition from non-traditional online bank players, which are winning outsized customer attention.
Advanced technology and management skills are required to develop business propositions that will succeed in the multi-channel environment. The management question that should be asked early and often is: “Which emerging business problems will benefit the most from big data’s competencies, given limited resources?” For many banks, the first place to look is managing the customer journey.
Darryl Demos is a Managing Director, and Kaushik Deka and Hank Israel are Directors in the New York office. They can be reached at email@example.com, firstname.lastname@example.org and email@example.com, respectively.