Smart Strategies Require Smarter KPIs

Digital processes, platforms, and predictive algorithms transform the strategic role and purpose of key performance indicators. KPIs are becoming measurably smarter, more dynamic, and more adaptive. This makes them more versatile and valuable for securing strategic advantage. For data-driven disruptors like Alibaba, Amazon, Airbnb, and Uber, KPIs don’t simply monitor enterprise success; they proactively drive it. This shift creates innovative opportunities for ambitious leadership.

To reap the benefits of these next-generation metrics, a new level of top management insight and oversight is required. This article identifies a virtuous cycle of critical success factors that determine effective KPI leadership.

Our research shows digitally sophisticated organizations have flipped traditional KPI purpose and processes inside out. Instead of seeing KPIs primarily as analytic outputs for humans, leading organizations increasingly use them as inputs for machines. That is, management relies on KPIs to train, tune, and optimize their machine learning models for business impact.

The organizational, operational, and cultural implications of this big flip are enormous. Smart KPIs literally learn to improve their performance and the performance of the organization. This emergent capability creates novel value-added relationships between management, metrics, and machines.

The most important insight: KPIs ― both individually and collectively ― become central organizing principles for leadership investment in data and decision-making. People, process, and technology are radically reorganized around metrics. Data and analytic priorities, as well as decision-making authority, are redefined and determined by smarter KPIs. As we concluded in a recent MIT Sloan Management Review article, your KPIs are your strategy; your strategy is your KPIs. For top-tier transformers, KPIs explicitly shape the strategic leadership dialogue and debate.

Customer churn offers the canonical KPI example of how virtual interdependencies between data and decision-making coevolve. According to Harvard Business Review, the cost of acquiring new customers can prove to be five to 25 times more expensive than keeping existing ones. Customer retention is key to sustaining cash flow and profitability. This holds particularly true for subscription businesses, notably software, financial services, and mobile telephony.

For these industries, reducing churn ― the KPI that tracks customers ending their relationship with a company over a particular time period ― is a strategic priority. Even determining that period ― a week, month, quarter, or year ― is itself a strategic choice. Almost without exception, significant changes in churn rates command immediate top management attention.

In big data and AI environments, however, understanding ex post facto churn no longer strategically suffices; organizations seek to predict churn to proactively prevent it. Making churn a more anticipatory and prescriptive KPI requires a virtuous cycle approach. In short, “learning from churning” makes the KPI smart.

Data governance is key. Distinctions must be drawn between churn presumed (a customer who simply stops engaging, that is, no more visits, purchases, etc.) versus churn absolute (the customer who explicitly closes an account or discontinues a service). Similarly, differences between reactive and prospective churn must be understood. Customers can leave after specific bad experiences, such as poor service or unexpected charges. Alternately, rival options might appear, or the service becomes less compelling for other reasons. This prospective or silent churn is typically more difficult to identify or predict.

But, clearly, organizations seeking to devise a smarter churn KPI need to correlate negative customer experiences with propensity to churn. They want to be able to capture and chart the gradual disengagement behaviors that reliably lead to disconnection. They’d likely analytically invest in identifying those clusters and segments at highest risk for departure. That requires that these companies know what data sets would make the best resources for scoring or ranking “likeliest to churn” customers to prioritize preemptive and preventive action.

But even these analytics don’t go far enough, as one global telecom provider discovered. The company (which I advised) developed and explored several churn prediction models. It quickly discovered that its analytics lacked any meaningful assessment of customer lifetime value (CLV) ― that is, the long-term revenue and profit potential of the customer. Potentially high-value customers were not measurably differentiated from customers who reliably switched providers in pursuit of the lowest possible price (or, almost as challenging, customers whose constant complaining and refund requests made them money losers). Aligning the CLV KPI with the predictive churn KPI completely transformed how the company assigned resources and designed interventions for customer retention.

Smarter metrics align the immediacy of situational awareness with longer term strategic aspiration. Process architectures achieve this by digitally linking KPIs, data, and decision-making into virtuous cycles. In effect, digital transformation empowers smarter KPIs; smarter KPIs compel digital transformation.

“The KPI Virtuous Cycle” illustrates the interplay and interdependencies that inform and improve KPI investment. For “born digital” and digitally transforming enterprises alike, each of these components is an explicit top management responsibility. Yet, how they interact, influence, and reinforce each other is constantly changing. Digital leaders ― like Amazon, Google, and Netflix ― appreciate, embrace, and relentlessly invest in this cycle.

Data governance is alternately defined as “the exercise of authority and control (planning, monitoring, and enforcement) over the management of data assets” and a “collection of practices and processes which help ensure the formal management of data assets within an organization.”

Regardless of definition, the essential point is that data is publicly perceived and managed as a valued asset. Top management ― Amazon’s Jeff Bezos, Netflix’s Reed Hastings, Alibaba’s (now retired) Jack Ma ― explicitly emphasize data’s importance as a strategic resource. “Smarter KPI” digital organizations govern data as the vital ingredient for measurement and accountability. Conversely, smarter KPI enterprise data governance frameworks are powerfully influenced and informed by their most strategic KPIs.

To be clear, these KPIs are not largely or primarily financial. Customer, supplier, channel, and partner data are integral to performance parameters. For example, self-declared customer-centric or customer-obsessed organizations put their data governance initiatives in the service of customer-focused KPIs, including Net Promoter Score (NPS) and customer lifetime value (CLV). Data governance here is a means and a mechanism to facilitate a KPI end.

For purposes here, data governance extends to and embraces analytics. That is, the purpose and quality of analytics ― be they crude regressions or classifications to the most sophisticated deep learning genres ― are dependent upon data quality, quantity, timeliness, and lineage, among other factors.

Decision rights identify what decisions need to be made to drive both the business and strategic alignment — who is explicitly involved in making them, as well as defining a framework for how they will be made and supported through operating processes and tools. The decision rights literature is extensive; its assorted frameworks, such as RACI (which stands for responsible, accountable, consulted, and informed) are constructively rigorous. But as Harvard Business School professor emeritus Michael Jensen despondently noted, “Allocating decision rights in ways that maximize organizational performance is an extraordinarily difficult and controversial management task.”

Smarter KPIs inherently require ever clearer articulations, delineations, and communications of decision rights. As a process becomes more automated, for example, when ― and why ― does a data-driven algorithm get the “right” to make a business decision instead of a human? If sophisticated predictive analytics suggests likely risks or desirable opportunities would significantly affect KPIs, what decision rights might people be granted to intervene appropriately?

Conversely, how should decision rights be defined and determined by organizations that want to empower their people and machines better? No easy or obvious answers exist.

Yet, in an era of smarter KPIs, machine learning, and intelligent automation, top management is constantly forced to revisit and rebalance decision rights allocations. At companies like Netflix and GoDaddy, for example, digital processes are frequently designed and tested with automated decision-making in mind. That is, they aspire to optimize KPIs by vesting decision rights in data-driven algorithms that, technically, learn faster, better, cheaper, and with more scalability than any human could. Indeed, the absence of human intervention often becomes an operational KPI.

At this point, the interrelated power and potential of the virtuous cycle dynamic should be compellingly obvious: As leadership strategically redefines KPIs, the need for different data sets and concomitant analytics to quantitatively support those redefinitions assume urgency. As more or better data and analytics become available through, say, partners or channels, should decision rights be reallocated to create the agility, responsiveness, and immediacy they might suggest?

What’s more, as new data sets and analytic techniques become available, established KPIs may need to become more dynamic. As people and algorithms become more capable, perhaps they should be empowered in ways that will meaningfully improve KPIs. Indeed, the very act of exercising decision rights and discretion may generate data that can ― and should ― change how performance is perceived and measured.

Bluntly, identifying the lags and latencies between KPIs, data, and decision ― how tightly or loosely coupled those elements may be, or how carefully or assiduously they are mapped and monitored ― matters enormously. Siloed decision-making would be anathema; there are no virtuous KPI cycles without cross-functional executive collaboration.

Indeed, the churn example described above highlights how neither data governance nor analytic insight should be divorced from decision rights in KPI virtuous cycle design. Retention offers to prospective high-value churners, for example, can be personalized or customized. They could be automated or humanized, depending upon what the analytics suggests. Customers not worth retaining could also be identified. The organization could analytically empower people and machines alike to intelligently manage the CLV-adjusted churn KPI according to the strategic aspirations of the business.

These same KPI virtuous cycle principles apply to the NPS and CLVs themselves. The transcendent insight here is that top management will increasingly have to decide how best to balance its KPI portfolio of virtuous cycles, considering:

• How can the business optimize CLV, NPS, churn, cash flow, and revenue growth, for example?
• How should nonfinancial KPIs learn from ― and with — financial KPIs?
• How can better data governance of financial data inform churn management?
• When should efforts to preserve NPS and customer satisfaction scores take precedence over sales?

The combination ― or fusion ― of KPI virtuous cycles and digital transformation has moved these questions from the speculative and rhetorical to the strategic and operational. For top management executives and their boards, strategic, smart, and strategically smart KPIs will be synonymous with strategically smart leadership.