IBM reveals that getting ROI from analytics isn't easy. Get the most out of your big data by following these four IBM best practices.
Last week, IBM presented its internal work with analytics and big data to analysts. The focus of IBM's presentation was its organizational journey on turning analytics into business value, and what it had learned from its journey. IBM focused on the challenges in turning big data into value-added intelligence as an example of how other organizations are also wrestling with getting the most out of their big data analytics.
"Data-based analytics can reveal client preferences, reduce operational costs and increase marketing effectiveness," said IBM in its report. "However, turning analytics into captured value (return on investment) is no easy task. Good data science is a necessary foundation for every analytics project, yet many projects still fail to achieve their full potential. Analytics value is not just derived from obtaining data volume and variety any more -- now critical value drivers are led by veracity (trustworthiness) and velocity (speed to action)."
IBM listed four key factors that lead to extracting maximum value from big data analytics:
- Priming the field by selecting data sources based on the potential for acceptance rather than initial perceived perfection;
- Easing the pain for users to understand and to work with this data by providing relevant insights that are easy for users to quickly understand and act upon;
- Going the distance to integrate analytics into the business-as-usual workflows; and
- Expecting improvement to big data and analytics processes with the incorporation of feedback mechanisms to cleanse data and foster new stages of future analytics.
Priming the field can be a political as well as a tactical process; it involves engaging with users to identify incoming data sources that users believe are reliable, accurate, and valuable for the business. However, a user in customer service might disagree with a user in manufacturing on which data is "best" for analytics on customer dissatisfaction, product returns, and product defects, especially if each department uses its own system and data. In this scenario, IT is usually the "data broker" that must get both sides to agree on where incoming analytics data is going to come from.
The next task, easing the pain, can be an even greater challenge. Users have high expectations for every minute they put into big data analytics projects, and they expect immediate returns on their effort investment. These returns must come in the form of actionable analytics and end user tools that are easy to use. IT historically underperforms when it comes to user friendliness in apps -- it can't afford to do this with analytics dashboards and drilldowns.
Easy-to-use tools ease the process of internal adoption, which can often be more people- than task-oriented. To integrate business analytics tools and functions into established operational workflow entails convincing users to change. These users have spreadsheets and reports that they are familiar with; they have even designed "workarounds" for the shortcomings of these reports, because they have become accustomed to the pain points. All of this adds up to a potential for user resistance to operational change, so the business value of analytics must be compelling -- in data equality/value and ease of use.
Finally, big data analytics must be delivered with an attitude and expectation of continuous improvement. It is human nature to just breathe a sigh of relief and say to yourself "I'm glad that project is over" and move on to the next project, but is the project really over? New enhancements to software are asked for as soon as a system goes live. Analytics is no different.
The warning from IBM and others is that, for analytics to stay relevant, it can't be treated as a finished project. Feedback mechanisms should be created to continue the analytics refinement conversation between IT and end users, with new improvements continuously being added.
Is this achievable?
Yes, this is achievable -- if IT as a big data "orchestrator" gives equal time to the people as well as the technical management of big data and analytics.
- IBM Watson: A shining example of how to take big data to the next level
- The big data challenge: Extracting actual business value
- Build big data adaptability through rapid experimentation
- Repurpose big data to get more analytics bang for your bucks
- How big data is going to feed nine billion people by 2050
- Research: 29% using IoT to collect big data (Tech Pro Research)
- The Power of IoT and Big Data (ZDNet/TechRepublic special feature)
Note: TechRepublic, ZDNet, and Tech Pro Research are CBS Interactive properties.