In a 2018 MIT Sloan Management Review (SMR) research report, “Leading With Next-Generation Key Performance Indicators,” Executive Editor David Kiron and MIT researcher, Michael Schrage, described the critical role of today’s KPIs in improving corporate leadership. Last month, the authors extended that discussion in an SMR article describing how machine learning (ML) and Artificial Intelligence (AI) can actually transform the impact and influence of strategic metrics.
The research strongly suggests that in an era of ML, enterprise strategy is defined by the KPIs leaders choose to optimize. Developing a strategy for AI is not enough, they say. Instead, actually using AI to create strategy will be at the core of algorithmic innovation in data-driven organizations. ML creates what the researchers call “the big flip” in KPI role and purpose. Instead of KPIs being outputs to inform human decision, we now see them increasingly becoming inputs to train machines. In turn, KPIs are datasets for machine learning.
Schrage, a research fellow at the Sloan School of Management’s Initiative on the Digital Economy (IDE), talked about the findings and their implications with IDE Editorial Content Manager, Paula Klein. Interview highlights are featured below.
IDE: The more you dive into the KPI rabbit hole, the more complex and multi-faceted the metrics seems to be. Is it both rewarding but also more complicated to be a data-driven enterprise? Let’s unpack that.
MDS: Complexity, simplicity, and data are in the eye of the beholder. Arguably, the clearest and most compelling takeaway from our research is that serious business strategists need to clearly articulate what they want their most desirable outcomes to be. Specifically, having a purpose and mission statement is nice, but what metrics and what numbers best capture your vision of success?
You can’t just rely on financial metrics or KPIs as your answers (as Robert Kaplan’s Balanced Scorecard anticipated over 25 years ago.) You need KPIs that make you distinct, distinctive, and compelling for your customers, employees, and investors. Which strategic KPIs best define and differentiate you? We increasingly have the data and computational resources not just to make KPIs more valuable, but to make them smarter, as well. The smartest KPIs will be the ones that learn.
IDE: What are the prerequisites and processes for strategic AI? Which comes first, figuring out and solidifying your KPIs or understanding the strategic value of AI for the business?
MDS: Our strategic premise is that AI, ML and KPIs are means and media to strategic ends — not ends in and of themselves. That is, we see AI, machine learning, and Key Performance Indicators as enablers, platforms, and springboards to achieve desired strategic results. Consequently, asking, ‘What do we want to achieve and accomplish?’ is the better strategic question and challenge than ‘What do these data-driven technologies and metrics allow us to do that we couldn’t do before?’ If you can truly describe how you want to transform customer expectations, build client loyalty, deliver greater value for less, become more agile and responsive, etc., then we can have meaningful conversations about designing — and prioritizing — the KPIs that best capture your strategic aspirations.
IDE: Why do you so ardently believe that “leadership teams that can’t clearly identify and justify their strategic KPI portfolios have no strategy?”
MDS: As for leadership teams, if the CEO, CFO, CMO, COO, and business unit leaders can’t tell me what their most important non-financial KPIs are and how they should be prioritized, I’m prepared to say they don’t have a meaningful strategy. How could they? Our point is simple: Serious strategists have the courage, integrity, and rigor to use strategic KPIs to hold themselves and their organizations accountable for measurable strategic outcomes. I don’t think that belief is controversial or provocative. Aren’t strategic KPIs exactly what a board of directors should be overseeing?
IDE: Organizationally, who will be involved in creating and using these metrics — perhaps a cross-functional team based on the goals? What are some of the different managerial skills you see emerging?
MDS: Those questions highlight the growing importance of quantitative leadership in a big data era. Strategic leadership, not operational or tactical management, needs to define the strategic KPI portfolio. Leadership needs to decide what kind of data, analytics, and algorithms best enable and inform strategic metrics and KPIs. Corporate leaders also need to identify and resolve inherent KPI conflicts. For example, are we maximizing individual customer transactions and sales, or are we optimizing ‘Customer Lifetime Value’? Are we looking to grow our most profitable customers and clients, or are we focused more on growing market share? There are no easy or obvious answers to these inherent KPI conflicts. So I’m forced to conclude that leadership becomes more important as strategic KPIs become more important. Besides openness to greater data-driven experimentation, the most important leadership skills will revolve around elicitation, facilitation, and collaboration for KPI improvement. Let’s call it ‘KPI kaizen.’
IDE: What’s needed from the AI/ML/algorithm technology perspective? Can you offer an example where carefully selected KPIs became AI’s strategic purpose? How can businesses shift to considering ML as a means to the ends of better business outcomes?
MDS: Anyone who looks seriously at AI, ML and KPIs instantly recognizes that strategic alignment between data, analytics, KPIs, and customer experience already exist in companies ranging from Alibaba to Amazon, Google to Netflix, Uber to Microsoft and Booking.com. That’s not an accident; it’s their strategic essence. For obvious reasons, leadership at these companies not only understands the power of platforms and ‘data as an asset,’ they’re committed to using innovative KPIs to spur new efficiencies and new value. Our Sloan Management Review article makes clear that marrying AI/ML to KPIs gives legacy organizations a fighting chance to accelerate their digital transformations and gaining keener insights into themselves and their customers. That’s a big win!
Much of the problem lies in culture and legacy. Traditionally, KPIs have been used as scorecards. They’ve been more about ‘how we’re doing’ than ‘where are we going’? Our research identifies how ML creates what we call “the big flip” in KPI role and purpose. Where KPIs were once outputs to inform human decision, we now see they’re increasingly becoming inputs to train ML algorithms and models. That is, KPIs are how digital processes literally learn to become smarter. We think that’s a big deal and it’s one of the reasons we believe this article is so important to organizations that really care about the future of strategy and the future of machine learning.