Representative End User Clients
Representative Automation Clients
Representative Software Clients
My recent research (Big Data for Industry) focused on how data from the Industrial Internet of Things (IIoT) will demand a re-architecting of the analytics infrastructure. Simply put, IIoT data will arrive so fast and in such volume that traditional approaches will not be able to offer up insights in a timely way. However, as some friendly feedback pointed out, that report really concentrated on anticipated (or expected) decisions. There's a different group of decisions - the unanticipated - which I didn't discuss at all.
On the left of the 2x2 matrix above lie anticipated decisions (or questions). These are decisions that the enterprise expects to make many times, at various frequencies. Strategic decisions sit at the top left of the model and are re-visited once or twice a year. Operational decisions (bottom left) are made much more frequently - perhaps daily, hourly, or even more often in some situations.
Because these questions and decisions are all anticipated, they are reasonable easy to support with business intelligence and analytics - at least in theory. (In practice though, many such projects underpinned by a data warehouse fail to address the users needs for various reasons). For many companies though, it is the unexpected and unanticipated questions on the right of the matrix that they truly struggle to cope with. (Note, the quadrant for unanticipated strategic decisions is blocked out because there really shouldn't be any...). To manage unanticipated decisions and find answers in a timely way, a complementary approach is required. In the world of relatively simple data and relationships, that role is best filled by visual data discovery tools. These solutions provide a rich interactive environment for users to explore data and seek out cause-and-effect. Where data is more complex - such as machine sensor data - predictive analytics may be required to tease out cause-and-effect relationships.
Either way, my advice is always the same: Plan your data and analytics infrastructure to support unanticipated questions, because I guarantee you'll get some.
On the left of the 2x2 matrix above lie anticipated decisions (or questions). These are decisions that the enterprise expects to make many times, at various frequencies. Strategic decisions sit at the top left of the model and are re-visited once or twice a year. Operational decisions (bottom left) are made much more frequently - perhaps daily, hourly, or even more often in some situations.
Because these questions and decisions are all anticipated, they are reasonable easy to support with business intelligence and analytics - at least in theory. (In practice though, many such projects underpinned by a data warehouse fail to address the users needs for various reasons). For many companies though, it is the unexpected and unanticipated questions on the right of the matrix that they truly struggle to cope with. (Note, the quadrant for unanticipated strategic decisions is blocked out because there really shouldn't be any...). To manage unanticipated decisions and find answers in a timely way, a complementary approach is required. In the world of relatively simple data and relationships, that role is best filled by visual data discovery tools. These solutions provide a rich interactive environment for users to explore data and seek out cause-and-effect. Where data is more complex - such as machine sensor data - predictive analytics may be required to tease out cause-and-effect relationships.
Either way, my advice is always the same: Plan your data and analytics infrastructure to support unanticipated questions, because I guarantee you'll get some.