Analytics Leaders: What You Can Learn from the BI Community’s Mistakes
Creativity and pervasiveness are competing drives in the realm of analytics:
- Creative analytics is about using data in novel ways to generate new insights and transform the business
- Pervasive analytics, on the other hand, emphasizes standardization and delivering a consistent view of metrics at scale.
This poses a dilemma for analytics leaders because, while it is difficult for both drives to coexist, both are what’s needed to make analytics successful. This is true for all types of analytics, and people analytics in particular.
Consider diversity as an example. Driving changes in diversity doesn’t work just by ordaining it from the top. Each manager must assess employees fairly, understand trends driving the underrepresentation of certain groups, and recognize his or her own personal biases. The supporting analytics need to be a part of every hiring and staffing decision for every manager. This is a classic use case for highly pervasive analytics.
But standardized and highly pervasive metrics often fail to incorporate new knowledge. For example, many business leaders used to assume that extroversion was a prerequisite for success in sales roles. But then academic research found that the correlation between extroversion and sales performance was pretty much zero. Those organizations that had the agility to quickly construct new metrics based on this insight were in a better position to open up an enormous pool of potentially better and cheaper hires.
So the imperative is to marry two the extremes — creativity and pervasiveness. To this end, it is important to understand how the business intelligence (BI) community has previously failed in this endeavor. This way, analytics, business, and IT leaders will not be doomed to repeat these mistakes.
The History of BI: An Oscillation Between Two Extremes
Over the course of my 30-year career in the BI industry, I have witnessed a continuous oscillation between creativity and pervasiveness. The following typical scenario shows how organizations fall into this counterproductive cycle:
The Analysis Tool Cycle — Creativity First
Company X has operated with centralized reporting and analysis capabilities for decades now. But then a few analysts — who are drawn to the creative side of using data to innovate — start using cool visualization and reporting tools to deliver unique, one-off insights. The company’s data scientists also start using analytics processing platforms that can take statistical algorithms and provide insight to specific challenges. People are, of course, excited by these new visual forms and the novelty of these insights.
In an attempt to capitalize and drive change, the organization adopts these new approaches and attempts to repeat these successes. As more and more people, functions, and groups adopt this new approach, the rate of innovation goes up, but alongside the innovation, they also start to see a growth in confusion. Metrics are defined differently across functions and even within departments. Two different groups within HR producing headcount numbers for the same employee population will produce different results.
The conversation then moves away from solving business problems and becomes fixated on whose data is better or right. Eventually, the organization shifts focus to the “Single Version of the Truth” and starts a process of standardization. The whole focus reverts to being about the pervasiveness and consistency.
But as the organization strives for standardization, the fact that the novel visualization tools were never conceived to solve that problem means it must be imposed by processes and rules. This feels bureaucratic and leaves everyone frustrated. It starts to override the very creativity that brought those tools to prominence.
The Analytical Storage Cycle–Scale First
As each analyst tool reaches a peak of inflated expectations, the storage side comes to the rescue, offering capital projects to construct new and more complete collections of data. This leads to an IT team taking a structural approach and planning to build some centralized and holistic warehouse. The challenge here is that the pace at which those assets can be constructed, loaded, cleansed, validated, and delivered is far too slow for an agile business environment.
Even when the warehouse is finally constructed, loaded and validated, new challenges arise. Complete warehouses are innately complex, requiring really sophisticated skills and knowledge to understand and leverage. To resolve this, the helpful IT team starts to construct simple, more domain-specific subsets in the form of data marts. As the analysts apply their creativity, each of these marts starts to diverge in definitions and loses touch with the “Single Version of the Truth.”
Then, the cycle flips and starts again.
Creativity and Pervasiveness: Drivers of BI Innovation
This endless cycle has driven multiple waves of technology innovation and vendor disruption. Business Objects (which is where I spent the bulk of my career and is now part of SAP), was the darling of creative analysts during the late 1990s and early 2000s.
Their original tool allowed the analysts to turn their database or data source into a rich semantic layer, create gorgeous views on that data, and deliver insight no one had ever seen before. As the pendulum shifted from creative to pervasive, however, the vendor bolted that technology into a server to manage all these generated views and semantic models. They were able to provide broad pervasive metrics consistently.
We can also see Tableau following the same pattern, moving from creativity to pervasiveness. Tableau grabbed the excitement of the analyst community at the end of the 2000s, offering intuitive analytical interaction and gorgeous visualizations. The company then rose to leadership in the following decade. As we can see with its recent Salesforce deal, Tableau is looking to complete the solution and provide reliable mechanisms to roll out hundreds of dashboards to thousands of individuals reliably.
In the realm of analytical storage, the inverse pattern has emerged, with vendors moving from standardization to creativity. This happened in the move from the MOLAP/ROLAP engines of the 1990s to the analytical models added to the core RDBMS platforms of the later 1990s and early 2000s, to the unstructured storage platforms based on Hadoop to the current Big Data offerings that leverage in-memory and are cloud based. The result, however, is the same cycle of technology change as tension arises between time-to-value and standardization.
A New Paradigm for People Analytics
Clearly, the need for consistency and governance is always competing with the need for creativity and agility. But delivering both at the same time is no longer impossible.
Instead of building a people analytics solution from scratch using traditional BI tools, now there is an option to pursue hosted, domain-specific analytic platforms. The most advanced platforms do not attempt to be just a tool for creative analysis, they also aim to be a point of consistency that can document every metric and provide secure distribution of well-articulated analysis. Built on a dynamic, in-memory analytic model, the best platforms ensure every element is reusable and connected — which allows best practices to emerge and be shared.
When these solutions serve a well-defined business domain, they are able to reconcile time-to-value with consistency. For example, in the best hosted people analytics solutions, voluntary turnover is defined according to HR best practices. This helps business and HR leaders avoid disputes over everyday metrics, and frees the data scientists and analysts to focus their efforts on the specific questions requiring a more creative approach.
Only a platform that’s geared towards HR will reflect the complexity of human dynamics. Classic BI was born out of a need from finance. Unlike financial data, people data follows a single object (the person) through hundreds or thousands of changes over a long period of time, which requires taking a really long, interactive view of the object to draw the connections. This is something that traditional BI tools just simply weren’t designed to do.
HR practitioners, analysts, and business leaders need a new approach that can unlock creativity without creating chaos. By learning from the mistakes of the BI community and making strategic technology choices, leaders can enable a culture of widespread, innovative decision-making based on a consistent view of data — all while removing the chains of process and policy.
Be the first to know!
Never miss a story! Get the Outsmart newsletter.