The democratization of data

Editor’s note: Diana Pohle is senior director, business analytics and insights, at biotechnology research company Myovant Sciences, Brisbane, Calif. 

On the surface, the democratization of data holds wide appeal for managers of the most innovative of start-ups and the stodgiest of incumbents alike. Few would deny the obvious advantages of accelerated decision-making and a culture of innovation that a data-driven orientation can nurture.

But the optimal data-driven operating model for any organization isn’t obvious. An organization planning for organic growth within an established footprint likely has very different needs from one planning for new vertical integration or an international expansion. Your confidentiality and cybersecurity needs, for instance, are very different when you’re sprinting toward a launch than when you’re undertaking merger and acquisition due diligence, or even when the target of due diligence is primarily invested in intellectual rather than physical property.

That is why an organization’s operating model itself needs to be every bit as nimble and flexible as any market-responsive activity undertaken by the organization. It needs to be as dynamic as the organization’s markets are, open to continuous reappraisal and adjustment in response to ever-changing business conditions and new market information. The model needs to be selected based on a dynamically weighted matrix comprising architecture, audience preferences and abilities, scalability and talent flows.

Where is your data? From centralization to sharing

Centralized, command-and-control systems are uncommon these days. Deeply siloed back offices that unify data collection, analysis and decision-making are largely limited to organizations with unusually large investments in physical assets. These systems may be justified by substantial barriers to entry, legacy investments or preoccupation with liability or confidentiality concerns.  

Much more common are federated or distributed services, which retain centralized authority alongside autonomous units that are equipped with their own in-house capabilities and usually distributed either regionally or according to disciplinary focus. A key advantage of federated services is a high level of specialization. Routinization of centralized services can lead to compelling efficiencies if good internal feedback mechanisms are nurtured. Like purely centralized services, however, federated services tend to enshrine and reinforce existing functional separations, imposing artificial boundaries and steep transactional costs between service delivery and user needs. Because user needs are dynamic but centralized workflows are typically static, distributed services tend to lag user needs and impose an artificial one-size-fits-all approach on all queries. User needs become disciplined by service imperatives rather than by market signals, thus depriving the organization of opportunities for learning and evolution.

The shared services model explicitly pools functions across multiple business units, encouraging multidisciplinary collaboration and enterprise-wide focus. It strives to recycle and repurpose data across many platforms and thereby achieve higher levels of customer service and performance management. When the shared service model is correctly implemented, it can pay compelling dividends, such as reduced time-to-information, the consolidation of duplicated services, an enhanced awareness across the organization of the value of data and its limitations and liabilities and, perhaps most importantly, stakeholder ownership in organizational risk and risk-taking. Unfortunately, all these gains can come at a price: standardization that can require lengthy pre-processing and result in long lead times and additional bureaucracy.

Both federated and shared service models entail serious risks. Centralized services may afford the organization a focus on security at the expense of nimbleness and responsiveness to market change, while the multinodal transparency and vulnerability of shared services can expose the organization to multiple points of attack. 

Insights leaders will benefit from weighing the strengths and risks of each model with respect to the goals of the insights function as they take action to update their sharing systems.

Focus on outcomes, not functions

More recent enhancements to the shared model seek to build on its strengths and mitigate its liabilities. Beyond merely democratizing access to data, these enhancements seek to cultivate resilience and adaptability by answering the question, “What does the cross-functional matrixed organization need to succeed in light of the accelerating velocity of data and the global distribution of flexible talent pools?”

First and foremost, this entails a new emphasis on outcomes rather than tasks. Rather than focusing on data collection, analysis and curation as discrete tasks, this collaborative services model instead focuses on how to apply and leverage insight. This is accomplished by building robust, interdisciplinary internal networks that are empowered and encouraged to crosstab market intelligence with the internal transactional data generated in its acquisition.

Every industry and every organization will have a unique constellation of opportunities to cultivate and deliver cross-functional insight. Much the way that organizations seek to provide seamless omnichannel experiences for external customers, established insights leaders may benefit from employing an internal omnichannel mind-set by architecting insights delivery to obscure the differences between analytics disciplines. 

Each organization will have different starting conditions, competencies and critical constraints, so each will need to develop its own road map for data-driven decision-making. But each will also require certain essential elements: topline focus, executive sponsorship, transparency and prioritizing the user experience. Consider the following:

  • Direct the focus of insights plans to the top line of the organization’s revenue statement or mission. Every department, program or project team –  even those not typically involved in data acquisition and analysis – can be encouraged to explore how every operational or tactical decision can drive incremental revenue or value in service of the mission over a more singular focus on project needs.
  • Engender support from an executive sponsor to accelerate transformation. Depending on the organization, this could be as simple as obtaining the endorsement of an influential executive sponsor or communicating the strategic linkages that justify dialogue about transformation. Vulnerabilities such as financial challenges, a low stock price, a challenging new product launch or an emerging competitive threat can be meaningful impetuses to secure buy-in.
  • Create transparency by de-siloing data. Spreading access to data across functional silos is a key step in naturally reaping richer insights. Take steps to ensure that insight is internally consistent and cascade it to appropriate stakeholders such that matrixed partners are banded together with a coherent brand story. 
  • Prioritize the internal customer experience.  Omnichannel best practices should not be limited to external customers. Internal customers expect insights leads to take the steps necessary to unify the analytical story across all touchpoints.

Democratizing data: Glimpsing the future with AI?

Though every organization will have its own opportunities and tools to manifest data-driven decision-making, one tantalizing possibility is the use of AI chatbots for internal platforms as a means of automating non-value-add activities.

Many organizations have used chatbots for elementary B2C functions for years, but few have begun to realize their potential for greater efficiencies in internal functions beyond addressing employees’ basic HR, payroll and benefits inquiries. Properly set up to classify user queries and responses, an AI chatbot could be used to not merely answer simple FAQs but also as a valuable tool to democratize data with expedient self-service access to data. The chatbot could ultimately be configured as a single source of truth for market research and analytics data, while addressing the ever-growing desire for real-time data at a velocity only dreamed of before. 

A transformed insights and analytics organization can operate with a surplus of insight. Its greatest challenge evolves from how to generate insight to the enviable new challenge of informing decision-making.