Supply Chain Report 2017: State of Analytics

Table of Contents
Introduction: State of the Industry
State of Analytics: Defining End to End Transparency (below)
State of Automation: Digitization is Not Digitalization

Today’s supply chain leaders have a problem. They are drowning in data, yet low on insights. In short, they cannot see. Data is everywhere, but not yet in useable formats.

The evolution of supply chain analytics is an exciting opportunity for business leaders. Traditional approaches focus on building analytics as an extension of existing enterprise applications: advanced planning, customer relationship management, enterprise resource planning, supplier relationship management, transportation management and warehouse management. New analytics approaches are designed based on data flows and requirements.

In this new world, data pools and streams inform business teams using new forms of analytics. The analytics infrastructure synchronizes, harmonizes and visualizes data flows. The analytics layer sits between the traditional application layer and workforce productivity.

These advancements are happening at both the enterprise and inter-enterprise level. The significant advances are being made in the following areas:

Cognitive Computing: Within five years, decision support technologies like supply chain planning, revenue management and sourcing will be transformed through the use of cognitive computing. This new approach will allow teams to drive new business value through semantic reasoning. The challenge for most teams will be trusting new forms of analytics. Companies will have to divorce themselves from spreadsheets.

Open Source Analytics: When e-commerce providers could not scale on relational database technologies, they evolved open source capabilities on Hadoop. Massive parallel processing enables schema on read handling and greater flexibility in the design of analytics. Instead of being stuck with inflexible hierarchies, schema on read enables the building of hierarchies and relationships by reading the data. This allows capabilities to evolve.

Machine Learning: Companies are plagued by disparate data. The view is that it’s “dirty” data. The reality is that most silos within an organization need data with a different context. But the traditional approach of hand-coding master data is expensive and outdated. The use of machine learning facilitates the read of data when needed and the shaping of data to the business context. Machine learning will transform master data processes within two years.

Streaming Data Architectures: Today’s businesses operate on batch processes and latent data. With the evolution of the Internet of Things and advanced sensors, new processes will evolve to facilitate decisions at the speed of business. This includes redefining replenishment from the outside in, redefining digital manufacturing to transform maintenance, and building new processes for service of heavy equipment, utilities, and asset-intensive operations.

Unstructured Data Mining: Within the organization, 70% of data is unstructured. This data is essential for understanding consumer sentiment, warranty and quality data. Mining this unstructured data enables its use for building a better understanding of how consumers view their products based on consumption patterns.

These advances are not an extension of existing processes and technologies. Testing and learning is required to understand the value and limitations of these technologies, which then will drive the design and implementation of new processes.

As shown in Figure 2, the evolution will be based on a confluence of new technologies. The continued dependency on spreadsheets is limiting the evolution of supply chain visibility (see Figure 3).