Going Beyond Data Science Toward an Analytics Ecosystem: Part 3
Understand the interaction of analytics environments for a thriving big data analytics ecosystem
Part 1 of this series introduced the emerging role of the data scientist and detailed some of the challenges organizations face when bringing this skillset onboard for their big data analytics initiatives. Part 2 then delved into the roles, responsibilities, and tools that compose the core analytics ecosystem. This concluding installment describes the extended and external analytics ecosystems.
Extended analytics ecosystem
The extended analytics ecosystem includes individuals and groups who use analytics functions. In organizations that have successfully applied analytics, this ecosystem may include most of the organization because analytics can benefit each of its functions. Like the core analytics ecosystem described in part 2 of this series, there are several key roles and elements involved.
Those with enough clout within the organization to enforce changes and the initiative to not only find but resolve problems are potential executive sponsors of big data analytics initiatives. Essentially, the higher the role of the sponsor, the more successful the program. Programs sponsored by the chief executive officer (CEO) are not unusual because the impact of analytics can potentially change the direction of the organization. However, more typical sponsors are the chief marketing officer (CMO), chief financial officer (CFO), or chief operating officer (COO). Cost-saving, financial risk, and waste reduction tend to attract the attention of the CFO; COOs may also be very keen on cost reduction as well as improved customer service. Many CMOs have already been using analytics to improve campaign management and customer segmentation.
The value of a given piece of information is not always immediately clear. Only by combining data from different sources can the value be made evident. For example, one of the key use cases for analytics is the assembly of a 360-degree view of customers from all their interactions with the business. However, sometimes data resides in disparate systems that are managed by different departments or lines of businesses, and combining this data, perhaps for the first time, may require obtaining buy-in from the owners.
Having a subject-matter expert (SME) in an analytics program is not a luxury; it is a necessity. Without it, the problem may be poorly framed and the insights gained may be erroneous.
The role of the analytics architect is to introduce tools, link them to the appropriate data, and harvest knowledge gained to make it available to other stakeholders. As Jay Liebowitz said, “Persuading users to put data at the heart of their day-to-day decision making takes more than just a megaphone about the virtues of data, or an order from senior management to use the data. For users to be ready to embrace the data in their roles, they need to be able to find it, understand it, and most of all, trust it.”11 A great deal of development has been undertaken to enable end users to experiment and discover insights on their own.
External analytics ecosystem
The external analytics ecosystem places the organization within the wider data supply chain. Although a natural first step for an organization is to start with its own data, incorporating data from outside can multiply the value generated. Furthermore, an organization can benefit from the big data analytics paradigm in other unconventional ways by, for example, offering its own data and analytics to other organizations.
External data providers
Many widely accept that the more data you have about customers, assets, or risks, for example, the more insights you can generate and thus hopefully make better business decisions. An emerging condition is that the data created and maintained outside of organizations is sometimes more important than the data that was acquired from internal sources. One reason behind this condition is that the resolution of external data is usually much higher than with internal data. Think of Twitter, Facebook, blogs, mobile phones, or radio frequency identification (RFID) data. These sources generate data about many events that are unlikely to be captured by internal systems. Open data initiatives, data marketplaces, and vendors who offer data sets are emerging in every aspect of business and society. Some of these providers such as Dun & Bradstreet, for example, have existed for a long time. Others such as Clear Story Data, Alteryx, or the World Bank are emerging in many areas of business and society. But how can this data be accessed? The answer is that the ecosystem should cater for that access by utilizing the appropriate application programming interfaces (APIs) to get to the data required—either to store it within the enterprise or to process it in real time. In either case, analytics architects should facilitate this access by using stream-processing tools, for example.
External data consumers
Making data available to external organizations is a key mechanism for business-to-business (B2B) relationships. During the last 30 years or so, much of this interchange has been automated. Big data analytics has, of course, increased the use of this electronic data exchange. However, an organization’s capability to monetize data is now possible not only by using it to improve the business performance but also by selling its internal data to external consumers. For example, effective data monetization enabled a large credit card acquirer to offer merchants additional value-added services such as analytics or report packages. Geolocation-based offers and location discounts, such as those offered by Facebook and Groupon, are examples of data monetization leveraging new emerging channels.
Cloud analytics platform and business analytics as services
One problem with big data analytics is that its workload is resource intensive. For organizations that do not have idle servers to throw at big data workloads, cloud-based analytics platform as a service (APaaS) can enable them to avoid the cost of implementing an in-house big data infrastructure. Pay-per-use options provide a good fit for big data analytics workloads where large resources are needed for a finite amount of time. However, one problem with a public cloud is the need to transfer huge amounts of data. Business analytics as a service (BAaaS) provides comprehensive analytics applications to its end users. An example is BlackRock, a large asset manager that makes its investment and risk management system—called Aladdin Investment Management System—available to over 150 pension funds, banks, and other institutions.12 Ironically, many of these institutions are competitors of BlackRock, which indicates an emerging pattern of business ecosystem that is fueled by coopetition—cooperative competition.
Big data analytics vendors and consultants
Vendors of big data analytics infrastructure, software, and services are an important part of the ecosystem and should be leveraged and integrated within the organizational ecosystem. Some vendors offer an extended version of BAaaS that enables organizations to derive value from analytics without the up-front investment in time, people, and technology. As with any outsourcing arrangement, determining a well-suited balance between outsourcing and insourcing is very important. The external analytics ecosystem includes other entities and groups that must be considered in plotting the analytics journey. Crowdsourcing is a valuable tool for performing experiments and gaining insights from a wide population. Regulators are also key stakeholders.
A thriving analytics ecosystem
Analytics is a game changer. It is revolutionizing how individuals, businesses, and society can use technology. Analytics is also flexible; it can be used merely to analyze limited data for a single task, or it can change the entire business landscape. The full value of analytics can be realized only when applied to integrated data from multiple sources and when insights are immediate and actionable. As with any organizational capability, analytics should be explored gradually to understand what value could be gained from it, but this exploration should be done in a way that enables it to grow so more value can be obtained. Viewing big data analytics as an ecosystem provides the understanding of how to chart the way to start small while enabling growth to achieve advanced levels of maturity and value. By observing the success or failure in building a big data analytics capability in small and large organizations, several recommendations can be adopted. Start small but tackle real business problems. Common sense dictates developing big data analytics capability in small steps. But this action does not mean applying analytics to an insignificant problem. A real business problem can demonstrate the value of analytics and pave the way to address other, perhaps bigger, business issues. Starting small could mean leveraging cloud computing platforms to avoid having to invest and set up a huge infrastructure. It could also mean using consultants instead of going through a hiring process that can take a significant amount of time and effort, and may not provide all the skills needed. Start from where the organization is now, and decide where you want it to be. Although big data analytics is a young discipline, it inherits components from a number of more established areas such as business intelligence (BI), information management, and enterprise data warehouses (EDWs). Hence, many organizations are not necessarily starting big data analytics from absolute zero. If they have capabilities in these areas, they can harvest some of their data, skills, and tools to start their journey. Conversely, not every organization needs to be at the level of Google or Facebook in its volume and capabilities. Each organization needs to decide where it wants to be and chart its course accordingly. For example, some organizations may not need their own big data analytics infrastructure and may be able to leverage cloud-based big data analytics such as APaaS or BAaaS. Enable connected data, insights, and actions. There are some key obstacles that prevent organizations from maximizing the value that can be derived from big data analytics. These obstacles are disconnected—fragmented, incomplete, and not integrated—data; disconnected insights (hunches and pet theories with no support by data); and disconnected actions (insights that are not actionable).13 From the outset, make sure the data-to-insights-to-actions loop is closed. Consider all the key roles of the core analytics ecosystem. As discussed in part 1 of this series, the data scientist role is crucial for a big data analytics program. However, if an organization neglects the data steward, analysis can be performed on the wrong data, security and privacy considerations can be compromised, or there may be many other undesired business risks and consequences. Without the analytics architect, organizations risk the disconnected data-insights-actions syndrome mentioned previously. In other words, organizations may end up with successful experiments that cannot be put into production or achieve the desired business outcomes. These recommendations, of course, neither comprise an exhaustive list nor a recipe for sure success, but are merely guidelines that can contribute to the triumph of any organization’s big data analytics endeavor.
A holistic view of the ecosystem
An exclusive and unbalanced focus on the role of a data scientist may not be the most effective way to establish a strategic, practical, and sustainable approach for big data analytics. Instead, organizations should view this capability as an ecosystem that consists of many elements that have to interact, collaborate, and integrate to provide real business value. Organizations considering big data analytics are encouraged to adopt this holistic view and to consider all success factors. And professionals in adjacent domains are invited to complement their skills and step up to fill the current shortage. Please share any thoughts or questions in the comments. 11 “Big Data and Business Analytics,” edited by Jay Liebowitz, CRC Press, April 2013. 12 “Financial Analytics as a Service,” by Ben Lorica, O’Reilly Strata Data blog, December 2013. 13 “Best Practice Guideline: Big Data,” Association for Data-driven Marketing and Advertising (ADMA), 2013.