With the adoption of AI in business processes, KPIs can provide a window into the health of an organization at granular levels, crystalizing the performance picture and removing ambiguity.
Becoming a “data-driven organization” has been the buzz for quite some time in the circles of the CIO, CTO, and now CDO, but there isn’t necessarily agreement as to what that actually looks like. Because the goals within each part of a business can vary greatly in expected outcomes, it’s often hard to determine whether a data-driven project was an overall success and actually moved the needle for a global, dynamic organization.
Much of the data supporting decisions about “did we do what we intended?” and “what should we do next?” are being generated by Key Performance Indicators (KPIs). KPIs are supposed to be about providing data that can be analyzed, reviewed, and acted upon to improve something about how the organization operates. Yet, a recent MIT Sloan survey of over 3,200 global executives stated, “Nearly 30 percent of respondents say their organization’s KPIs only somewhat, minimally, or not at all drive how they lead or manage their people and processes. A larger portion, more than 40 percent, say their organization’s KPIs are moderately influential.”
KPIs have been around a long time—what once were relatively humble transactional benchmarks have evolved into complex measures of behavioral patterns and performance. With the adoption of machine learning and AI in business processes, KPIs can potentially provide a window into the health of an organization at granular levels, crystalizing the performance picture and removing ambiguity.
See also: 8 Digital Transformations Coming in 2020
But, there’s a flip side (a dark side?) to this capability: just because you can measure something doesn’t mean you should, or that it has anything to do with the health of your organization. As used by many organizations today, KPIs may be valuable within functional areas of an organization, provided they are actually measuring results that really matter. The problem is, those results aren’t necessarily tied to the bigger picture—the strategic vision for the organization. This disconnect has the potential to derail an organization’s progress and send decision makers off in the wrong direction and resolving it should be a top priority. But how to achieve that will take rethinking our relationship with data.
Feeding the (Analytics) Beast
We all know the cycle: KPIs create data that feed analytics, analytics are used to make decisions, KPIs measure the impact of those decisions. And, of course, the results are only as good as the data. Today, organizations are generating and capturing more data than they know what to do with. Identifying the right, relevant data, analyzing it, then using it to guide the organization are key functions of leadership.
To drive better decision-making, organizations apply analysis techniques in ways that can predict results (“predictive analytics”) and recommend actions (“prescriptive analytics”). Now, to be fair, this is nothing new. Forecasting and choosing which path to follow are basic functions of every kind of organization. What’s changed is the technology that supports these kinds of analyses, making some aspects of it easier, faster, and more comprehensive than ever.
Thanks to the expanded availability and cost-effectiveness of AI tools and processing power, the modeling required to accurately predict outcomes is continually improving. The automation and machine learning components of these tools can deal with higher volumes of data while providing more nuance, since they’re designed to reflect the human judgment of what is important and relevant to the organization’s strategy. Assuming the inputs are valid, and the queries are presented properly, predictive analytics will look at what has happened and spell out what should or could happen next.
Of course, it’s up to leadership to decide if these predictions really matter and, if they do, what steps should be taken. Most business intelligence tools in the market today stop at the point of predicting the likelihood of events occurring as a result of actions already taken. That’s why there is so much interest in prescriptive analytics, which lays out what to do with the results of the predictive step.
Prescriptive analytics is a more subtle science. It’s a methodology that takes the results of predictive analysis to chart a course toward a goal, either by establishing rules for decision-making or by laying out paths to optimize processes that support the goal. As more organizations look for ways to apply the mountains of data they collect to reaching strategic targets, many are looking to this approach. In fact, while approximately 11 percent of large and mid-size organizations are using some form of prescriptive analytics, Gartner forecasts this will grow to 37 percent by 2022.1
But the more encouraging trend is in seamlessly combining these two disciplines, as organizations try to maximize value from their data to propel better decision making and, obviously, results. According to Gartner, “By 2020, predictive and prescriptive analytics will attract 40% of enterprises’ new investment in business intelligence and analytics.”2 Still, while some industries (manufacturing, pharma, financial services, logistics) have been using predictive and prescriptive analytics for some time, it’s usually in siloed parts of the business, as opposed to being applied globally.
As with so much of organizational decision-making, leaders need to avoid letting the tail wag the dog. It’s critical to start with the big picture, strategic vision of what the organization wants to achieve. Every decision, at every level, should always tie back to supporting that goal.
So, what does all this have to do with KPIs? If KPIs are established in a vacuum (i.e., “What’s good for this one department or process?”), then what’s the likelihood of them helping to reach the top level, organization-wide goal? Remember, KPIs are supposed to be measurements, not targets. With data in silos, and the KPIs that feed the data sets also being siloed within functional units, there’s nowhere near enough correlation between what’s going on in various parts of the organization and what’s supposed to be happening to support the health, growth, and sustainability of the enterprise.
The challenge, then, is aligning ‘what do we know?’ with ‘what should we do next?’ To resolve this, leadership must be committed to gathering and analyzing only information that is relevant to the strategic goals coherently and across silos, in an orchestrated way.
The Case for Orchestrated KPIs (OKPIs)
The modern analytics market is rapidly expanding and evolving, with tools to peer into every aspect of an organization’s processes. But that glut of options prompts the challenge of orchestrating all those tools — a task that can be more costly and time consuming than acquiring the tools to begin with. Analytics orchestration, of course, is what we do to harmonize the collection and analysis of all this disparate analytical data — regardless of where in the enterprise those analytics are churning out insights.
What if we took that same approach with KPIs?
What if — just as companies need an “orchestration layer” to make sense of diverse algorithmic processes and business contexts — we crafted an “orchestration layer” of KPIs (OKPIs)? These would be cross-function and cross-departmental KPIs that sit on top of the business unit KPIs to interpret and align their meaning. The result is a more holistic and unified understanding of their significance and value to business impacts and outcomes. It will unify leaders within the organization to act as one cohesive unit, rather than sales operating at a different pace than marketing, or finance looking at different markers than operations. From there, success becomes a distinct and measurable number, rather than an ambiguous and relative outcome that benefits bits and pieces of the organization. Once this top-down clarity is achieved, leaders can begin to set KPIs for their team, based not only on their individual goals, but the company’s goals.
OKPIs are by no means hypothetical. Startup and bootstrapping environments, in particular, are increasingly leveraging what we could consider OKPIs to stay agile and avoid waste from misaligned metrics. The result is a better understanding of the performance of their solutions, more accurate market data and more realistic milestones. Having all these KPIs orchestrated ultimately reaps a far better understanding your products and markets.
Think “KPIs by Design” for More Standardized and Stronger Data
Earlier, I described the OKPI layer as a place where organizational KPIs “sit on top” of the departmental KPIs. That’s true, but don’t confuse that for some shallow layer application that keeps data ambiguous or forces you to guess at what’s meaningful deeper into the stack. Not unlike the Secure by Design movement, KPI’s should be baked in—refined and standardized as early in the development process as possible. That comes with full buy-in from the executive team and all leaders within the organization.
The earlier those KPIs are codified, the more their many variants will have in common as we build out our analytic architectures. Otherwise, they simply reflect the individual experiences of the users, failing to align them against the common, orchestrated KPIs. The result is a lot less coordination… and insight. By contrast, the more proactively and strategically we embed KPIs into our processes and architectures, the more progress toward an intended result.
Hitting a target is only useful if that target needs to be hit, and all too often, employees are going after a metric that is only marginally related to moving the ball forward. Unfortunately, within some organizations, it’s become more important to achieve the KPI metric than to improve on strategic priorities. That leads to wasted effort, and certainly doesn’t improve either employee engagement or the health of the enterprise.
All this, of course, is founded on a data-driven culture. We strike out to set clear goals and expected outcomes — always the first steps when developing a data and analytics strategy. But in order to do so, a company must first share KPIs across the organization that everyone’s working towards along a seamless continuum of individual, team and company goals. OKPIs are what align all the other, subordinate KPIs on the continuum, and the result is strategic and operational improvement, a stronger analytical basis for decision making and an enhanced focus on what matters most.
1 Forecast Snapshot: Prescriptive Analytics Software, Worldwide, 2019,” Gartner, January 2019
2 Combine Predictive and Prescriptive Techniques to Solve Business Problems,” Gartner, October 2018
