Most businesses are primarily concerned with customers, internal operations, suppliers and product. They are not primarily interested in business analytics. So it seems fairly obvious that if analytics is to have a place in the business then it needs to address these things – customer recruitment and retention, supply chain, internal efficiencies and product based activities. It is also the case that operational activity in most businesses is defined by various processes, and within these processes are hundreds if not thousands of decision points. If analytics can help make these decisions more effective through the provision of risk reducing information, then it will be performing a very important role. In fact most businesses are mostly defined by just three things – processes, transactions and decisions. We’ve largely automated the first two, and so it is time for analytics to provide assistance and possibly automate decisions.
Production analytics puts decision enhancing analytics at the point of work, and there is almost no part of the organization that cannot benefit. Reduction of customer churn, customer credit limits, fraud detection, predictive maintenance of machinery, optimized staff recruitment, up-sell and cross-sell offers, supply chain optimization and a thousand other processes that are heavy with decisions can benefit from analytics. But the analytics need to be embedded into the production applications. Having to switch from one toolset to another destroys productivity and causes errors. So production analytics is largely embedded analytics, helping people work more effectively and efficiently while using their everyday applications.
The use of analytics outside the production environment is always experimental. If it wasn’t then it would be embedded into the production environment. We do of course need people to explore data, but the organization as a whole cannot become one large analytics research and development enterprise. This will have obvious effects on productivity, because it is not part of the production effort.
Ultimately nearly all analytics will become production analytics – a convergence of different analytical methods and production applications. Visual analytics, predictive analytics, machine learning, prescriptive analytics, simulation and even AI will be embedded into the production environment with the single purpose of either assisting or automating many decision processes. The current fascination with BI as an activity that is divorced from everyday operations is a fashion that will probably disappear as quickly as it appeared. Suppliers who are best positioned for this move are those who provide true embedded capabilities and integration with diverse analytical methods. GoodData, Sisense and Logi Analytics come to mind. Once BI suppliers catch the scent of production analytics on the wind then we will see a wholesale movement to integration of analytical methods, and the delivery of analytics at the point of work.