The global data challenge
To read the full article, learn about JOS Recommendation and receive updates, please register.
The emergence of citizen data scientists
One way that enterprises are embracing big data is extending the role of data analytics. Until recently, it was primarily the job of the data analysts and data scientists. But, these days there are new players in the game – business users. Or, as they are now being called, “citizen data scientists.” They are usually power-users who know the business domain intimately and understand the meaning of data.
Most existing big data tools provide only operational reports to enable these power-users to monitor historical or real-time business performance. With the right data architecture, building hypothesis reports for ad-hoc and predictive analytics to exploit new opportunities is now a real possibility. Business users can do many of the things a trained data scientist typically does, without needing to learn programming languages like Python or R.
“IT teams tend to want to standardise on a single solution,” said Ty. “That isn’t the best thing for data science, because it can limit the number of people who can get involved in the analytics process.”
Citizen data scientists don’t happen by accident. Companies needs to develop a corporate culture that values data and ignite the analytical mindset among all business users. Only if they are motivated and encouraged, users can extend the capabilities of interpreting historical data for business insights to building ad-hoc hypothesis reports for exploring new opportunities.
“An open mind and a willingness to embrace a variety of tools encourages more extensive exploration and predictive modelling, and helps to spread a data-centric culture to every corner of the organisation," he added.
Empower insights with data lake
Choosing the right tool, or mix of tools, is important for building the architecture and culture for user-driven analytics.
Today, most analytics dashboards are driven by data warehouses. The bad news is that it’s difficult for a traditional data warehouse to digest the flood of information streaming in from different sources, such as IoT devices or social media platforms. It also limits business users to build hypothesis reports with the data to explore new insights. The good news is that data lakes are quickly changing that picture.
“With a data lake, you can throw everything into a single repository and immediately start looking at the information, without needing the IT department to do a lot of work setting things up,” said Ty.
But this does not mean data warehouse becomes obsolete. It is still the best tool to manage structured data and support operational analysis. Data lakes could complement the architecture by offloading workload at the data warehouse. Its key strength is to easily accommodate the fastest growing segment of the data landscape – unstructured data.
“In the digital age, data is the new oil. And, like oil, data needs to be mined, processed and applied to different applications. Data lakes enable enterprises to acquire, blend, integrate and converge any-and-all types of data, regardless of source or format, to support business users developing their own analytics algorithms," Ty said.
Data lakes are also more flexible and scalable as businesses grow.
“When you create a data lake with the Hadoop open-source software framework for storing data and run applications on clusters of commodity hardware, you can scale out in a way that until recently only a Tencent or Alibaba-sized business could imagine,” he added.
Cracking the coffee code
When it comes to big data, seeing is believing.
With a culture for citizen data scientist, Starbucks Coffee Asia Pacific recently worked with JOS to explore new business insights from its newest and most up-market venture – Starbucks Reserve Bars. The new outlets featuring special coffee beverage and brewing methods offer what APAC Finance Director Brian Liu describes as “an elevated coffee experience.”
Initially, Starbucks' existing financial system was not catered for the new business model at the Reserve Bars, which produce more customised and specialty beverages. This made tracking performance a cumbersome and largely manual process. However, after implementing an analytics solution based on Tableau, the contrast is sharper than between a café latte and a grande americano!
According to Liu, financial reports that used to take days to produce can now be done in minutes. The finance department also saves time from data gathering and focus more on analysis. This encourages more business end-users within Starbucks Coffee Asia Pacific to become more curious about the data and explore it themselves.
“It’s now pretty easy to identify areas for improvement in Reserve Bars. We can also do ad-hoc analysis, and build other dashboards or worksheets to see what is really going on,” said Liu.
There are more big-data projects in the pipeline. Starbucks plans to migrate cross-functional teams, like HR, supply chain and store development to the platform, so it can enhance functional capabilities and get more insight from data. Eventually, Starbucks hopes to create a host of different dashboards to form a single coherent picture of the company.
“Being able to leverage the full power of big data allows us to make better decisions, identify more business opportunities and move ourselves to a more sustainable business practice. And, with JOS and Tableau, we should have more time to enjoy our coffee,” said Liu.
Achieving success one byte at a time
Managing the 3Vs — volume, velocity and variety — is no longer enough. Today, making a big data strategy a reality means also creating genuine business value through data veracity.
Mature organisations that already have a culture of user-driven analytics can quickly set goals and find tools to achieve big-data ROI. But, for others, identifying and articulating such precise details is not always easy.
Despite being a daunting initiative, a user-driven analytics strategy should start sooner than later. “The sooner you start, the further ahead you’ll be. An effective enterprise-wide analytic makes a difference between success and survival,” said Ty.
He recommended avoiding the risk of confidence-shaking big-data failures by starting small with relatively simple predictive modelling projects, then rolling-out proof-of-concept solutions to build on that success.
“Tackle simple predictive analytics projects, then build on that success with more sophisticated use cases,” he suggested. “Working with a third-party expert can also help enterprises understand where they are today, and what they need to do to achieve their long-term goals.”
A successful big data strategy is based on more than having the right infrastructure in place. It’s also about how to expand capabilities economically and enable users at every level of the business to conduct analysis that help to drive brightest possible future.
How do you feel about empowering user-driven analytics in your organisation?
How do you feel about building a data lake in your organisation?