"When you take something rules-based, you are not able to adapt predictions to new data," said Mike Flannagan, SAP's senior vice president for analytics, ahead of the company's Sapphire Now customer conference in Orlando.
"The power of machine learning is you are able to continually update the model. Your model is running against all the data it has seen so far."
But there's another stumbling block to that: the computing power required for machine learning systems. "Most business apps aren't robust enough to handle the machine learning computation," said Flannagan. S4/Hana, on the other hand, is fast enough to embed machine-learning prediction in a core ERP system, something that was previously only possible with rules-based prediction, he said.
That technology shift is a game changer when it comes to tasks like using the sales pipeline to predict the next quarter's revenue. "You don't have to wait till you have closed the books," he said.
