TDWI | Training & Research | Business Intelligence, Analytics, Big Data, Data Warehousing

Data Strategies for High-Value Analytics

The outcome and effectiveness of any analytics application will be influenced by the volume, breadth, sources, models, quality, and semantics of the data provisioned for it.

• By Philip Russom
• May 22, 2020

Many businesses today are expanding their analytics programs because they know that analytics is an established way to achieve important business goals, such as revenue lift, increased competitiveness, customer retention, customer account growth, new product development, operational efficiency, and cost containment. To get the greatest return on investment (ROI) from each new analytics initiative or implementation, these same organizations are also extending the data integration infrastructure that provides data for analytics. After all, the old adage "garbage in, garbage out" still applies, even with modern analytics.

For Further Reading: 5 Steps to Creating a Data Strategy to Drive Your Nonprofit's Social Good A Case for Managing Data Uniquely for Each Form of Advanced Analytics How to Make Better Data Decisions

The outcome and effectiveness of any analytics application will be influenced by the volume, breadth, sources, models, quality, and semantics of the data provisioned for it. Furthermore, data requirements for analytics vary across the many forms of analytics, such as methods based on reporting, dashboards, statistics, data mining, text mining, natural language processing, graph, clustering, neural net, predictive modeling, machine learning, and so on. To achieve the fullest ROI and shortest time to use for analytics, organizations must employ data integration and other data management techniques and tools to provide just the right data, in just the right format and condition, on a per-use-case basis. In short, you need a data strategy -- with plans, people, skills, and tools to back it up -- for each analytics application.

How does an effective data strategy increase the ROI of analytics?

Analytics professionals must thoughtfully align their solutions with business goals, and data management professionals must provision data that is tailored to specific analytics solutions. Both of these demand some kind of strategy. Otherwise, analytics is weak, poorly targeted, and of minimal business value and ROI. To understand this better, let's drill into strategy, data strategy, and analytics ROI.

What is a data strategy?

Data aside, Webster's New World College Dictionary defines strategy as "the science of ... maneuvering forces into the most advantageous position prior to actual engagement." This definition was penned for a military context. Yet, it applies very well to data integration, which maneuvers and combines data assets into the most advantageous models and aggregations prior to using the repurposed data to engage with customers, partners, business processes, and analytics. Strategy can also involve a plan for action and skill in managing and planning. In that spirit, a data strategy is usually documented in a plan that will guide the integration and management of data for a specific purpose such as analytics. Furthermore, the best data strategies can flexibly apply to individual analytics applications as well as long-term, multiphase analytics programs.

What forms does data strategy commonly take in the context of analytics?

Enterprise data standards are a strategy for managing data. The standards prescribe metrics for the quality of data in most data sets, methods for modeling data in certain contexts, and the preferred interfaces for data in motion.

Data governance policies are a strategy for assuring regulatory compliance and data privacy. A policy specifies which users or user roles can access which data sets, domains, and applications.

A policy may also specify preferred techniques for protecting sensitive data (such as encryption and masking) or methodologies for executing governance (such as stewardship and curation).

Technical data strategies relative to data platforms can be driven by data's structure or use case. For example, relational data for analytics is typically managed by a relational database management system (RDBMS), whereas file-based, large volume, raw source, and unstructured data may be managed by Hadoop. As another example, TDWI sees many organizations storing analytics data for reporting, dashboards, and OLAP in a traditional data warehouse on premises, whereas data for advanced forms of analytics is progressively managed on public clouds. Such preferences can constitute a data strategy.

For what business and technology situations do you need strategies focused on analytics ROI?

In an ideal world, business managers will determine corporate goals first, then IT and other technical staff will create strategies to support the business goals. For example, when the business expresses a need for better customer retention, data management professionals may capture and prepare behavior data and other customer data so it can be analyzed to predict customer churn and action can be taken to prevent it. A coordinated effort such as this—involving business management, data, and analytics—typically has a demonstrable ROI in the form of continued revenue from retained customers.

As another example, when the business announces a need to demonstrate its compliance with a particular regulation, data management professionals may tighten up and better document how access to sensitive data is controlled, as well as how analytics is conducted without regulatory infractions. The result is a "soft ROI" in the form of assured compliance, along with a "hard ROI" in the form of fines and sanctions avoided.

Note that some data is time sensitive, such that its business value decreases as the data ages. For example, an e-commerce recommendation engine needs a data solution that can capture website visitor behavior in real time, then pass it to an analytics solution (probably with predictive modeling embedded) to do real-time analytics to determine a compelling recommendation. As another example, high-speed manufacturing needs the analysis of real-time data from the shop floor to assure the fulfillment of the day's service-level agreement relative to unit count. In addition, real-time data and analytics are key to business value and ROI in utility firms, trading exchanges, logistic services, transportation, and other industries. These use cases and industries need special data strategies that enable the capture and analysis of real-time data.

In summary, many CEOs and other chief officers want to better unify their enterprises and enable more collaboration among colleagues. This can be supported by standardizing and sharing quality data across employees and departments, as well as providing analytics capabilities to more employees. Depending on the details of execution, sharing data and increasing analytics can lead to ROI in the form of greater efficiency and innovation via knowledge sharing.

For Further Learning

Much of this article was excerpted from the TDWI Checklist Report, Data Strategies for Accelerating the ROI of Analytics. Download this report to read more about strategies for high-value analytics.