Today's Big Data deployments are missing the mark

Lyndsay Wise, president and founder, WiseAnalytics
Lyndsay Wise, president and founder, WiseAnalytics

President and founder of WiseAnalytics. Lyndsay Wise has a decade’s worth of IT experience, consulting with SMBs on business systems analysis, software selection and the implementation of enterprise applications. She also conducts research into leading technologies, market trends, BI products and vendors and data visualization — and, happily, has agreed to share some of her thoughts with visitors to We look forward to more from Lyndsay on the world of Big Data and related topics.

Big data implementations are becoming more commonplace as organizations struggle to address business challenges resulting from a lack of visibility into their data. Companies are being tasked with managing their information assets and figuring out ways to derive value from disparate data sources, both internal and external to the organization. This work requires data management strategies that tie into action-oriented business results, and that provide the basis for many big data projects. The reality, however, is that Big Data alone represents the tip of the iceberg for businesses that want to develop and maintain a strong data management program tied to quantitative business value. This article looks at how independent Big Data initiatives fall short, while offering insight into what steps organizations can take to tie their Big Data initiatives to quantitative business value.

Understanding the value of data

The value of data comes from an organization’s ability to act. Big Data on its own might help support a business’ aim to increase competitive edge through data acquisition and analysis, but to provide broader value, it needs to be tied to an organization’s strategic goals. After all, what can a big data project do on its own if it isn’t tied to taking data and transforming it into something that can be used to improve efficiencies or increase revenue?

Case in point. Competing products and services have an increasingly difficult time maintaining key differentiations due to the rapid and continuous advance of technology. When one company or brand develops a new product or service, competitors will follow suit within a short period of time. The only way to stay ahead of the competition is to develop strong brand recognition, high customer satisfaction levels, continuous R&D, and in-depth insights into data. This is where the potential of big data implementations exists. Unfortunately, many organizations stop at the creation of a big data platform and remain unsure of how to bridge the gap between data management and actionable insights.

 A practical look at where big data falls short

The telecommunications industry offers a good example of how and why internal, external, and social media data sources can help to ensure better service. Generally, most Internet, phone and cable services offer similar products with similar pricing. Promotions get customers in the door with promises of discounts for bundled services or percentage discounts for a certain number of months. Since many services are alike, perceived differentiation comes from levels of service, how complaints are managed, and the provider’s ability to customize products to meet individual customer preferences.

Doing this effectively requires more than big data. When someone complains on Twitter or calls in to customer service, there is the potential to address issues effectively and rectify the problem. There is also the potential to become vulnerable to customer churn. The only way to successfully address these issues is to analyze customer profiling, including demographics, sales, accounts receivable, service history, social media interactions, etc., and to combine this information to identify what can be done to rectify the situation. This one example applies to many vertical markets, such as retail, financial services, banking, insurance, and health care. Organizations that do leverage Big Data effectively tie its adoption to broader analytics, creating algorithms, for example, to identify the potential for customer churn, levels of current satisfaction, and the lifetime value of customers in order to identify which ones should take priority.

Unfortunately, many Big Data initiatives still fall short as organizations struggle to tie information to these types of business initiatives. To remedy this, closed loop processes are required. Understanding customer access points, how data is collected and what sources are required to develop analytics that can provide the right insight may help organizations transition from being reactive (sending letters to customers cancelling services) to proactive (flagging potential defectors and identifying ways to encourage them to retain current services). However, many organizations require a shift in thinking about their data assets and the ability to integrate new technology to support broader data access for analytics and process improvements.

First steps to attain business value from Big Data

Although Big Data platforms are becoming more popular within organizations, many still struggle with their value proposition. The first steps to tying data to business value are as follows:

  1. Identify Big Data infrastructure, platform and access point: Organizations should evaluate what data infrastructure will work best within their current IT architecture and identify the key object of their Big Data implementation. Looking at Big Data’s purpose, and at how it will be leveraged can help IT determine whether Hadoop, Cloudera, or some other platform will best meet business needs. The platform selected should create a stable environment that enables successful data management over time. Irrespective of how data will be leveraged, organizations should also consider data access, profiling, and cleansing so that trust and reliability can be maintained over time and not simply with the initial data load.
  2. Incorporate analytics to leverage data: While not all Big Data projects are tied to analytics, the reality is that to identify gaps in performance and gain quantitative insights into business operations, analytics need to be tied to Big Data deployments. This means evaluating the relevant stored data and creating joins so that analytics can be developed. In many cases, this will involve consolidating multiple data sources, including those that fall outside Big Data stores.
  3. Define desired outcomes: Predefined analytics outcomes can help an organization better manage their operations and identify performance improvement. However, issues are not always defined and businesses many need the flexibility to analyze information on an ad hoc basis. Doing so effectively requires two types of understanding: the first involves the ability to identify what needs to be analyzed on a regular basis (i.e. predefined analytics which can include customer related algorithms as described above), and the second requires self-service access to delve deeper into analytics as more questions arise. In some cases, this might entail direct access to data and in others it might require the development of easy-to-interact-with applications.
  4.  Aligning people and processes: Many organizations overlook the value of people and processes as they focus exclusively on the data side of the equation, rather than on developing closed-loop processes that can be tied to action. Understanding why customers act as they do, or identifying an issue within the supply chain only provides the initial phase of getting value out of data. Unless an organization identifies the right people and processes required to take action, to fix problems and to take advantage of opportunities, Big Data will not meet its potential value. This is why it becomes important to align processes, people, and technology to enable businesses to get the most out of their technology investments.

 Taking these four steps into consideration can help organizations transition from Big Data deployment towards a more holistic approach that ties management of information assets to closed-loop processes that enable data driven actionable results.


Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.