Data democratization demands good governance
- Posted on November 11, 2019
- Estimated reading time 5 minutes
![data democratization](/-/media/images/blogs/avanade-insights/data-democratization-demands-good-governance.jpg?la=en&vs=3&w=1120&hash=2D3DB54850BABD440CF81C2DA68315B6)
This article was originally published in Retail Customer Experience.
In C-suites across the world a comedy and a tragedy are playing out that are hindering good decision making. The CMO steps up to present with a beautiful set of charts built by his data science team. The CFO objects that the calculations of revenue in the CMO’s presentation don’t align with her own. The COO looks sheepishly at his primitive spreadsheet extracts while the GM for a major division looks peevishly at the CIO, who hasn’t yet prioritized her new reports and keeps trying to avoid eye contact. There is a huge missed opportunity for the enterprise as the C-team all have such separate lenses on “what is truth.”
Data democratization
We live in a golden age of analytics with the costs to collect and store data dropping rapidly. At the same time many products have been introduced that allow businesspeople a greater ability to access and visualize this data to make a decision or to communicate an important idea. This trend, data democratization, is to be encouraged and embraced. Data democratization represents a great cultural shift from gut-feel decision making, enabled by leaders like Microsoft with Power BI, which leverages many of the skills business users developed over the past two decades using Excel. However, poorly executed, democratization potentially gives rise to self-selection of data that simply reinforces gut-feel decisions.
In the best hands these types of tools elevate the level of C-suite communication and decision-making. If this standard isn’t consistently used by all divisions and functions, however, it can lead to confusion or an arms race for fancier visualizations. In the hands of an inexperienced analytics and visualization team, it can mean that data is misrepresented, bad decisions are reached and faith in the decision-making processes of the company is diminished.
The role of the CIO
In some companies the CIO has taken a leadership role and set the pace for analytics adoption. In far too many, however, he or she has either been outpaced by stakeholders’ ability to build teams and adopt tools or, worse yet, has gained a reputation as an obstacle to be overcome.
Companies lose a great deal when their CIOs aren’t central to their analytics strategy and adoption. As the keeper of the company’s systems of record, the CIO and his or her team are ideally positioned to play the role of data steward over the most granular enterprise information, ensuring that data sources are valid and that data models are used consistently across the business. The CIO is also uniquely positioned to create a community of interest or a center of excellence where data analysts and data scientists can share approaches and develop their skills. These organizational constructs can veer between physical and virtual, permanent and transient, but a guiding hand is required to help the business and IT understand their purposes, objectives and operating model.
Data governance
To fully resolve the dilemma of democratization - getting the benefits of the trend without a descent into chaos - companies need to take a step back and carefully reconsider their holistic data governance strategies.
There are five key areas that need examining:
- Funding and business cases. Data-centric decision-making done with best practices is a cultural shift for many organizations. There needs to be an implicit or explicit understanding that everything from front-line to executive decisions are better when supported by data
- Data model. If everyone is going to agree on the definition of truth, that truth needs certification, guarantees put around its integrity, and the truth often needs explaining, and through governance certifying, so that the overall data literacy of the organization rises. This requires a management lifecycle comprised of data creation, storage, movement, usage and retirement.
- Process model. There are two aspects to this: first, a process by which the business gains new capabilities to examine, model and analyze data; second, an explicit linkage between the data and the operational and strategic processes it supports.
- Experience. More than a matter of stylistic preferences, the way in which information is presented needs to convey something to all stakeholders and have a tight linkage to process so that the business can see the human and organizational impact of decisions quickly. This area of governance addresses the user experience, including information architecture, graphic design, and interactivity.
- Security and profiles. It’s tempting in a democracy to throw the gates of security open and avoid the issues surrounding the privacy, integrity and visibility of data – especially since some of the data will be in aggregate form and not identify a transaction. However, without well-address security profiles, the context and business focus around a decision can be lost, data can be compromised, and indecision and imprecision can result.
One of our clients, a global leader in the oil and gas industry, learned that not addressing these issues early in its development of a broad BI/Analytics program led to incomplete take-up of new dashboards and expensive re-implementation of competing non-correlated views by business area. Our team strengthened the linkage between hierarchies of metrics, developed easy to understand visualizations with clarity of business meaning and causation between metrics, and used change management to roll out new capabilities to ensure management approval. Results included a strengthened role of the CIO, improved responsiveness and reduced cost in managing centralized metrics, while increasing the accountability for decentralized metrics.
Enterprises are surprisingly diverse on some or all these topics. Some are forthright on their ability to achieve any form of consensus around these topics; others are in denial about the state of readiness to define even the simplest interaction. This makes governance, or trust and alignment development, an ideal opportunity to look at neutral third parties as a way of creating platforms for discussion of these topics – and eventually driving agreement to a set of definitions and rules without overhead.
Readers might want to seek out a neutral party with extensive data governance, optimization and presentation experience who can broker these conversations between business and IT to support the enterprise as it builds its next-generation operating model – and get started by:
- Identifying whether these democratization issues exist in their own organizations, to what extent and whether these are intentional or not.
- Understanding the role of the CIO and how he or she is engaged in the analytics space as a responder to requests or as a provider of direction and thought leadership.
- Looking to whether the business is ready to adopt a more formal governance model in exchange for quality data sets and improved experience.
Comments