Mastering the critical steps to being a data-driven organization
The organizational view of data has evolved from often being an afterthought to that of a fundamental currency that drives decisions.
A key element in this data evolution has been the exponential growth in technology that has enabled organizations to better aggregate, sift through, understand and reason with data. These abilities have transformed organizations from being “data aggregators” to “data-based decision makers.” Indeed, more than 75 percent of organizations plan to move to data-driven decisions by 2020.
To realize the potential benefits of data-driven decisions, many organizations need to transform itself from the inside out. To become a true data-driven organization (DDO) it is not sufficient to just invest, purchase and implement tools. Organizations need to embark on a journey from being a “data collector” to “data aware” to “data-driven” in a timely and strategic fashion.
To become data aware, organizations should focus on the right tools and technologies to refine their data architecture. This enables the understanding, extraction and integration of actionable data relevant to the business. Once data capabilities are enabled in this “data aware” stage, organizations should focus on data governance and data quality. In parallel, data architecture is refined and scaled to include analytics, predictive modelling, big data and artificial Intelligence.
On this journey of becoming data-driven, we see many organizations fail or stagger. As a result, 58 percent of today’s businesses still make at least half of their business decisions based on gut feel or experience instead of by leveraging data. Primary reasons for this stem from a lack of quality data, poor data governance, siloed data management, inflexible data architecture or lack of proper orchestration of multiple source systems.
A lack of alignment or communication on critical data systems and projects among senior leadership also hastens failure. Ill-equipped organizations often undermine and de-incentivize data capabilities. Eventually they fail to achieve the much-needed culture change of being data-driven and resort to old habits that fail to deliver desired results.
Data analytics has moved from being a support capability to a core competency with key analytics resources managed in an integrated model. Our engagement observations have shown that data-driven retailers are optimizing operations and leading their markets in various areas.
For example, data-driven retailers are forecasting inventory and managing supply variation based on environmental factors, maximizing profits based on price experimentation, reducing churn, enhancing personalization, tracking customer and product together, optimizing customer and product affinity optimization, optimizing trade through omni-channels, and more.
An organization can transform into a DDO by adopting different models. In evaluating their fit for an appropriate model, organizations should select the model that works culturally and structurally for their business. In recent years, a few core flavors of DDO have emerged including (but not limited to) the following:
- Integrated Model: A standalone function within existing IT organization reporting to the CIO.
- Parallel Model: A parallel structure outside of the current IT setup supporting the entire organization reporting to CXO/CEO.
· IT retains policy & governance control
· A steady transformation driven by IT who remain closest to organization data
· Data function reports to IT and does not have authority to enforce, override IT policies and get burdened by it
· IT has the authority to influence data function’s priorities and needs
· Data function acts as an independent organization and is not controlled by other functions thus maintaining
· Has the authority to influence IT and other functions in enforcing policies
· Data function gets undermined as vanity project that solely disrupts Business as usual
· Occasionally, lack of cooperation from IT and alignment challenges
Our experience with large, complex multi business units and multi-format retail organizations has helped us uncover that several retailers start out with an integrated model and gradually shift to parallel model and in some cases a hybrid model that capitalize on positive aspects of both constructs. In doing so, it allows for policy setting by IT as well as promoting independence desired by other functions or business units. Our clients in retail, consumer packaged goods, consumer electronics, e-commerce and retail finance have made advances on data and analytics maturity curve through use of structures outlined above.
In evaluating independence, authority, and degree of separation, the Parallel model best suits organizations’ transformation aspirations. Organizations that leverage Parallel model are able to embark on the transformation while maintaining separation for business as usual (BAU).
The path to being data-driven starts with identifying and assigning accountability to a chief data officer. The CDO will run the organization with the right model (integrated or parallel) to ensure that data is an integral aspect of every design and architectural decision. The CXOs should enable the chief data officer to drive changes independently.
To enable the transformation in a parallel model, the CDO should:
- Communicate Purpose: Socialize and get buy-in for parallel model by communicating rationale while owning accountability
- Set up Organization: Establish a Data function by analyzing existing organization for internal roles and hiring externally for roles that are not intrinsic to traditional IT.
- Data Analysts: To collect data, create dashboards and perform analysis.
- Data Scientists: To explore, transform, correlate and interpret data.
- Machine Learning Engineers: To write and draft algorithms to perform predictive analysis and enable the organization to learn from the data.
- Data Architects: To liaise between business and machine learning engineers to interpret business needs and context and to help engineers create appropriate solutions.
An effective data strategy includes hiring for these roles. Such outreach:
- Underscores commitment.
- Avoids prejudices from existing processes that employees are exposed to in large organizations.
- Incentivizes adoption among current organization to adopt to organization’s direction.
- Improves collective intelligence by importing external intelligence and expertise.
It is important, that data organizations have foundational capabilities covering six dimensions across the data value chain, as described below:
- Eliminate Silos: Traditional organization setup enable functional silos environments that do not incentivize knowledge sharing. Data analytics encourages the exchange of information, so CDOs need to devise policies and controls that identify, address, and eliminate silos.
- Enable Access: Data management and governance often result in intricate authorization and access policies that prevent lower reaches of the organization from accessing and visualizing insightful data. CDO should encourage open access and exchange of data across the organization so data can be converted into information and knowledge.
- Become Predictive: Most organizations are still stuck utilizing data for descriptive purposes (explore, report and analyze). To become a true data-driven organization, CDOs should incorporate predictive (decide, predict and act) and cognitive (learn, reason, understand and share) aspects of data along with descriptive purposes.
Questions can arise on what are the implications on the operating model. Core tenants of the operating model that need to be considered include the elimination of rigid organizational hierarchy, the progressive incorporation of data tools including cloud computing, the rationalization of old systems, the incorporation of new systems and trends, upskilling of the organization with training, and new culture initiatives and enhancement of in-house data capabilities.
Data transformation comes with risks and pitfalls. An independent data organization needs to attain credibility and authority in order enable success. The following steps help minimize or eliminate risk:
- Design and implement robust data strategy: Build vision, mission and strategic data goals for the organization and ensure that the strategy meets enterprise goals.
- Get Executive Sponsorship: Find and align executive sponsors on the strategy, key projects and milestones. Ensure that the culture flows top-down.
- Enhance data governance: Define processes and support structure with clear roles of accountability.
- Build Future-proof data architecture: Ensure that the data architecture is streamlined, flexible and scalable to support the existing and future needs of organization.
- Quick Wins: Implementing a quick win to demonstrate that new data capabilities are easy to obtain.
- Avoid Turf War: The CIO should enable a transformative culture by working in concert with CDO to achieve organization goals.
- Enable innovation: Cognitive and predictive data analysis involve multiple data sources (IoT, sensors, spatial). CIOs need to ensure that innovation labs are setup and appropriate research funds are apportioned.
The road to being a data-driven organization transforms the enterprise in terms of culture, capability, skills and ability to meet future demands, but the journey is often fraught with failures and missteps due to misalignment in intent and/or fit. Leaders who are challenged or daunted by the scope of transformation need to realize the risks and respond accordingly.
A successful data driven organization gets the three pillars of transformation right – people, process and technology. This powerful triumvirate – when done right - transforms the DNA of an organization and enables the enterprise to realize business value through data omniscience.
(The authors thank Anshuman Jaiswal and Vinodh Natarajan for their contributions to this article).