Image courtesy of Shutterstock.
Businesses need to be able to lead with technology, Edd Dumbill, vice president, strategy, Silicon Valley Data Science, observed in the opening keynote at Data Summit 2015 at the New York Hilton Midtown. To fully exploit the opportunity of big data tools and architectures, a new way of thinking is required, one that frames data as raw material. And, to transform data into value, IT must move from thinking about what it does to data, and instead focus on business outcomes and what can be done with the data to advance the business, according to Dumbill.
In the last 5 years, there has been a change as far as what IT does, moving from a focus on fast processing to a period where the potential for IT to create value in organization is much greater, and as a result IT is concentrating on helping the business reach customers, said Dumbill.
“Now, we can touch customers one to one.” The other new capability made possible by big data the ability for "any company to march into any business that they want, and this is really frightening for a lot of people,” said Dumbill. “Who would have thought that it would Netflix and Amazon commissioning the most interesting and exciting new TV series coming out?”
Today, more than ever, it is important to connect technology and business value. IT must consider how to articulate how the cost of the data systems relates to the benefits to the business; and articulate how the technology philosophy enables business aspirations. Any big data project, said Dumbill, has to start with business people because that is where it will be judged. IT can’t succeed without business and business can’t succeed without IT, he emphasized. Big data projects succeed when there is a project with a goal that is defined, and can be accomplished in a tight time window. While a roadmap must be defined in advance, you must also know there will be bumps in the road. Just as an artist needs to understand their materials, the business needs to understand what the technology can do and when the business has articulated what it needs, that is when IT can really get going, he said.
Increasingly, he noted, IT organizations need to take a lesson from the Silicon Valley giants such as Uber, Amazon, and Netflix, and learn from what they do. What characterizes these companies is their ability to learn. While everyone has heard the mantra of moving fast and failing often, we mainly hear about winners. By failing fast we learn to ask better questions. But we also need to be able ask questions quickly and take action quickly in order for big data to matter to the business for a feedback loop.
With big data applications, there is more uncertainty but the business potential makes it worth going for it. Data lakes offer a new approach with Hadoop for an elastic, scalable repository that is very different as an overall architecture than the ones we have had previously.
But in order to move fast, and fail often, organizations need to make failure cheap and make operations a platform for innovation rather than siloing off the people with ideas away from operations. In any data project, the most important thing you can do is identify data and bring it in, and of all the Vs in big data, variety is the one that holds the most value because that is what provides insight, Dumbill noted.
Data science is what enables us to efficiently take advantage of the fact that everything is digitizing, and to be responsive in the market.
Conventional data strategies are about what is done to the data, things like cleansing, validation, protection. Legacy data systems rely on predictable, manageable, and data; with applications tightly bound, and the database designed around application assumptions.
However, a modern data strategy allows companies to attract new customers, target VIP customers, and automate. A modern data strategy is about creating value for the enterprise, Dumbill said.
To access the slides from Dumbill’s presentation, go to http://conferences.infotoday.com/documents/235/0900_Dumbill.pdf.