The evidence of significant competitive advantage for early adopters of Big Data continues to come in. McKinsey’s report on Big Data covers a number of industry sectors. McKinsey believes there is as much as a 60% increase in retailer’s operating margins possible with Big Data. ” Across sectors, we expect to see value accruing to leading users of big data at the expense of laggards, a trend for which the emerging evidence is growing stronger.” http://www.mckinsey.com/insights/business_technology/big_data_the_next_frontier_for_innovation.
I’ve been focusing on Big Data the past few years and have visited dozens of clients. I believe Four Barriers today are preventing customers from this value, they are:
1. Value of Big Data Use Cases
2. Hot/Warm/Cold storage
3. Data Wrangling
4. Widespread Predictive Analytics
In future blogs, I’ll cover points 2-4 in more detail. For Hot/Warm/Cold storage, I very much like SAP’s Smart Data Access solution for both Federating and optimizing access to a variety of data stores, as well as SAP’s new embrace of HADOOP. I’ll also cover Data Wrangling, in particular, the need to supplement traditional Datawarehouse ETL processes. I’m also a huge believer in the value of Predictive Analytics, with its fact-based approach, but also strongly believe we can’t rely on the <1% community of truly qualified Data Scientists.
The right approach begins and ends with value. There are two general approaches to looking for value; tops-down and bottoms-up. A tops-down approach starts with executive strategies & goals, looks at the As-Is state and potential To-Be states, considers Data Architecture dependencies, develops a Roadmap of Use Cases and Value, and identifies Quick Wins. The bottoms-up approach starts with analysts and data scientists playing around with data, seeing what comes of it, and sharing the larger organization. Data Visualization tools, or Data Scientists running Predictive Analytics against a HADOOP cluster can be bottoms-up approaches. I’m a huge fan of both, but it’s best to know where you’re heading when you set out on a journey. As DJ Patil, Data Scientist and author of ‘Data Jujitsu’ says: “Before investing in a big effort, you need to answer one simple question: Does anyone want or need your product? http://radar.oreilly.com/2012/07/data-jujitsu.html. Start tops-down, very fast follow with bottoms-up.
Estimating value of Big Data Use cases requires three skills: Business Strategy and ROI model expertise, Big Data Use Case ideas with creativity for the art of the possible, and Big Data Architecture expertise. It’s common to have conversations that circle around Value, Use Cases, and Data Architecture. To break out of this endless loop, you should have Three Maps; a Value Map, a Big Data Use Case Roadmap, and a Data Architecture map. You need someone who has a strong working knowledge of all 3 areas. Just as finding a Data Scientist with Statistical, Programming and Visualization skills is difficult, finding a Big Data Monetization is equally challenging.
If you don’t manage all Three Maps, your planning sessions will loop around the topics. An executive might see big value in a Use Case for Price Optimization, but if you don’t know the best tools for this, and what data is required, your effort will stall. Similarly, your IT team might have some great ideas of Data Visualizations and Predictive Analytics of Manufacturing Yield information, but if you can’t estimate the ROI, your project will stall. The Three Maps need to be linked. This is the path to getting a quick win now and grabbing strategic advantage.
I will follow up soon on the next 3 barriers, and the solutions I’ve seen in the marketplace that overcome them. It’s time for leaders to accrue significant value over laggards.
VP Customer Innovation