Sergey Nivens - Fotolia
When Time Warner Cable wanted to gain deeper insight into the cable TV preferences of its 15 million subscribers, the media giant turned to the science of cloud-based analytics as a service. "We needed to get a better understanding of customers' behaviors within different demographic segments of our audience," said Jeff Henshaw, TWC's senior director of digital marketing and analytics.
With its ability to quickly scrutinize huge volumes of data, analytics has quickly become critical to strategic decision making. Consequently, it's no surprise that businesses are revving up spending to leverage the technology. In a November 2015 study, research firm IDC said worldwide spending on analytics services is expected to jump from $58.6 billion in 2015 to $101.9 billion in 2019 -- a compound annual growth rate (CAGR) of 14.7%. Writing in the study, Ali Zaidi, IDC's research manager for IT consulting and systems integration services, pointed to two key factors driving that growth: adoption of new technologies and a shortage of in-house analytics expertise.
TWC's challenge was to understand which cable TV packages its customers preferred based on age, income, geography, family size, presence of children and other factors. The company also wanted to correlate that information with the online behavior of its customers -- whether they use TWC's smartphone app or a browser app, follow their navigation paths from page to page, tally how long they linger on individual screens, note whether they complete a purchase and even identify where users were if they chose to abandon their sessions. The goal, Henshaw said, was to obtain answers to specific business questions and get to the heart of data-driven targeted marketing. In turn, that would help TWC adjust the product mix within its various cable TV packages, launch promotions, improve app navigation and even reduce telephone calls to its customer services representatives.
Four types of analytics
Data-driven cloud analytics as a service (AaaS) is often divided into four types, spanning a value spectrum that moves from hindsight to foresight.
- Descriptive analytics, the simplest to implement, addresses the "what happened" question and provides rearview-mirror hindsight into past activities. According to CI&T, a global IT consultancy headquartered in Brazil, about 35% of companies it surveyed do descriptive analytics on a consistent basis.
- Diagnostic analytics, one step up in terms of both implementation effort and value, provides for a deeper dive into past data to address the "why did it happen" question. In its own tutorial on analytics types, IT distributor Ingram Micro said one common use is aggregation of the varied aspects of a social media marketing campaign into a view that shows what worked successfully in the past. According to the CI&T's survey, less than 5% of companies perform diagnostic analytics consistently.
- Predictive analytics, the next step up, shifts from pure hindsight into a blend insight and foresight, answering the question "what might happen." Use the big data that has been amassed, correlate with other available and appropriate data -- such as weather, geographic preferences or economic data -- and predictive analytics has the ability to predict future data. "You can never predict with certainty what will happen, only what might happen," said Judith Hurwitz, president of Hurwitz & Associates, a Needham, Mass., IT consultancy. Fewer than 1% of companies in the CI&T survey are using predictive analytics.
- Prescriptive analytics is the ultimate, offering the highest value, but at the cost of complexity of implementation. It attempts to answer the question of "what should we do" by providing multiple options for dealing with business situations. To differentiate predictive and prescriptive, research firm Gartner said predictive analytics can forecast when a machine might break down, while prescriptive might suggest preventative maintenance actions. Though implementations of prescriptive analytics are uncommon, IDC, in its Worldwide Big Data and Analytics 2016 Predictions, said that by 2020, fully half of all business analytics software will incorporate prescriptive capabilities.
Where does TWC fall across this spectrum? "We use all four analytics types in some capacity," Henshaw said. "If I had to choose, I would go with prescriptive." By getting to what he called "data-driven targeting," the company's analytics implementation, a cloud-based analytics as a service from Adobe Analytics, provides TWC with suggestions for fine-tuning its offerings. Those adjustments eventually feed back in, creating a continuous process analysis.
Simon Jamesglobal lead for performance analytics, SapientNitro
The analytics as a service movement is spreading rapidly. In its December 2015 forecast, Research and Markets cited lower costs of implementation, ease of customization and agility as factors in its prediction that the global AaaS market will grow from $5.9 billion in 2015 to $22.24 billion in 2020 -- an impressive CAGR of 30.4%.
SapientNitro, the digital subsidiary of marketing consulting firm Sapient, is also a user of AaaS. The goal is to generate new insights from collected data, allowing the firm to recognize new marketing opportunities for its clients. "Analytics is an art form that's driven by conjecture," said Simon James, global lead for performance analytics at SapientNitro. "Analytics is a means to an end, either to better performance or improved gross profit margins. You want more certainty of success, but every answer inevitably leads to more questions."
Data meets analytics, but where?
Similar to other IT systems that are moving from on-premises installations to the cloud, analytics is no different. Given that big data often already is cloud resident, that's where the analytics should reside, too, according to Nik Rouda, senior analyst for big data and analytics at the market research firm, Enterprise Strategy Group Inc., in Milford, Mass. "It's natural to bring the analytics to the data. You don't bring the data to the analytics; that would be slow and expensive." TWC stores its data on the servers of the Adobe Analytics service.
Jim Comfort, general manager of cloud services at IBM, agreed. "It's one thing to simply store data in the cloud, but it's the analytics in the cloud that make data useful," he said. "If you need one, two or 20 different analytics approaches, you can easily do all of that with the flexibility and agility that a cloud services environment offers."
Speed is critical
For TWC, the main reason for implementing cloud-based was to speed business decisions. "We needed our implementation to be nimble, mirror what the business wants to measure and measure against a certain set of goals -- that's top of mind." For developers, that meant using dynamic tag management, embedding tags or pixels on pages. Using this method frees engineers from an iterative development cycle as business requirements change. "Tag management is very good way to get around that," Henshaw said.
It's easy for analytics to become an afterthought in the midst of efforts to get new business initiatives out the door quickly. Henshaw warned this should not be allowed to happen. "You have to guard against analytics getting pushed aside. As deployments go out, be sure your analytics are not broken and that tags fire properly." To do that, TWC walks new engineers through the inner workings of the entire analytics process. For developers, they need only know where to place to analytics code within each page and how to code to avoid latency."
Big data checklist for analytics project managers
Don't forget IoT when implementing analytics
Avoid these traps when deploying predictive analytics