Predictive Monthly: How A Little Bit Of Wizardry Can Transform Your Bottom Line

Pierre Leroux

Just think what Harry Potter might have done if he hadn’t spent his youth stalking Hogwarts’ dark corridors. He could have used his magical powers to build outstanding companies and become the bespectacled wunderkind of Wall Street.

Today the world is being reshaped by data, and businesses are separating out into wizards and muggles – those with insightful super powers and those without. Predicting the future is becoming the most sought-after superpower of all because it provides a chance to change the future before it happens. How different would your business be today if you could have spotted equipment failures before they happened, identified unhappy employees before they quit, or known in advance which customers would upsell and which wouldn’t?

Predicting the future from patterns in data

Since the advent of digitization and the Internet of Things, companies have been amassing data at an exponential rate: data from sensors in manufacturing and transport equipment, from social media, and customer interactions. Predictive analytics software makes sense of all this data through algorithms that find patterns in large volumes of data, which in turn enable predictions to be made about the future.

Predictive analytics is not entirely new; in financial services it underpins computer-assisted stock trading. In retail, predictions of online buyer behavior are increasingly being used to design personalized marketing campaigns. Elsewhere, companies that understand the competitive edge that it can provide are starting to invest. A recent Forrester survey of data and analytics decision makers found that that 39% have data and analytics budgets of at least $10 million. Furthermore, 52% are spending at least 5 percent more on advanced analytics technologies this year than last.

What makes predictive analytics so powerful?

  • Objective and accurate predictions: Unlike business intelligence tools that rely on human input to infer cause and effect between results, predictive analytics uses techniques called machine learning that teach the computer to look at a particular outcome and then uncover the factors behind it (which could include thousands of possible causes and nonlinear relationships). The result is far more accurate predictive outcomes that can improve over time. A leading German car manufacturer is using predictive analytics in this way on the factory floor; processing 30,000 data sets per second from up to 100 engine components to predict engine failure as early as possible – minimizing lost hours and resources.
  • Automated decision making: Algorithms are at their most powerful when used to automate decision making between two or more business processes (requiring no human input at all). A common example is a credit card algorithm that can lock an account if it spots that a purchasing pattern is abnormal for the cardholder. In manufacturing, where networked factory floors are being further networked into the supply chain, the adoption rate of predictive analytics is expected to expand rapidly. The 2016 MHI Annual Industry Report anticipates adoption rates reaching 80% (from today’s 22%) over the next six years. By digitizing analysis and decision making in this way, businesses are responding to situations in real time, achieving groundbreaking efficiencies.
  • New opportunities: Predictive analytics is also enabling organizations to discover new business opportunities and build digital models around them. An earlier Digitalist blog, Unlock Your Digital Super Powers, revealed how an airfare prediction firm in the United States turned its “waste” data to gold. By mining its archive of itineraries to correlate past trends with current fares, it can now find its customers the lowest possible price 95% of the time. In sales, a leading computer networking company is targeting $1 billion in extra revenues thanks to predictive analytics providing insight into what its customers are most likely to buy.

Making it happen

Developing models used to be a forbidding task and the sole domain of magical data scientists. Complex and meticulous, the data had to be well prepared and organized, algorithms wisely chosen, and models revisited regularly in case the assumptions behind them needed updating. With the latest advances in predictive analytics, however, lines of business can adopt an assembly-line approach and use automated techniques to achieve advantages such as:

  • The use of predictive models and machine learning can be expanded to thousands of business processes.
  • With access to guided workflows and automated techniques, the pool of people able to contribute to the predictive value chain is widened to analysts, smart data analysts, and citizen data scientists.

Read more about success with algorithms in this earlier Digitalist blog: Algorithms: The New Means of Production.

Many who invest in predictive analytics see it as an ROI decision rather than a cost because of the incredible value it can release from existing data and infrastructure. In my blogs over the the coming weeks, I’ll be diving deeper into how predictive analytics is bringing its magic to HR, sales, and manufacturing.

To learn how your business could embrace predictive analytics read Predictive Monthly: The Role of Predictive in Your Analytics Strategy.

Pierre Leroux

About Pierre Leroux

Pierre Leroux is the Director of Predictive Analytics Product Marketing at SAP. His areas of specialty include Data Discovery, Business Intelligence, Cloud applications, Customer Relationship Management (CRM), and ERP.