Very useful article by Mark van Rijmenam to which I will add:-

First- Most BI/analytics deployments based on looking in rearview mirror so just like driving a car look ahead to anticipate threats and opportunities withpredictive analytics

Second- exploit algorithm libraries, machine learning & AI but also combine these withself-service BI.This allows operational and line-of-business people to add intuition and local knowledge. A vital component of decision-making and sharing enterprise-wide competencies

Third- it is no use if BI and Analytics is deployed to only some of the organisation's people. They must give meaning and insight to everyone who needs to make better decisions. It is not for a select band of analysts, to be limited to specific departments. 

Fourth- it is not just about visualisation and infographics. It is about embedding BIin the fabric of an organisation's ERP, CRM, HR, Supply Chain, PMO etc etc. 

Then you have thereal platform for competitive advantage. The answer to counter the uberization of everything. The way to combine disruptive insurtech, fintech and othertech platforms without disruption