< Back to Home

Touchless Forecasting: The Future of Decisions in Demand Planning

Cognitive automation enables touchless forecasting that is faster and more accurate than manual processes.

There’s a new pace for all domains of business in the era of COVID-19, but especially demand planning. Chemical businesses in particular were left scrambling to change from annual or monthly cadences to weekly or even multiple daily ones in order to respond in real-time to market signals and drive executive decisions.

Top performers in the industry are now using cognitive automation, intelligent automation, and artificial intelligence across all domains as they adjust to this “next normal.”

Touchless Forecasting

The process begins with the creation of a Demand Planning Control Tower, where a series of end-to-end automation processes fuel real-time algorithmic planning.

Cognitive data layer: automated data streams harmonize endogenous and exogenous data in real time
Pre-processing: the data is cleansed and qualitative data is encoded
Forecasting: using a portfolio of methods, the system uses statistics and machine learning to make predictions
Ranking algorithm: the best metric is determined and top ranks are returned to the end user for advanced parameter tuning
Post-processing: the action approved by the end user is fed back into the system to fine-tune automation rules and improve forecast accuracy

As the system is used, the organization can use resultant metrics to identify what can remain truly touchless and where human intervention is a value-add as opposed to a necessity

Chemical Industry Case Study

This particular company is a Top 10 Player looking to improve Global Demand Forecasting. They had been using a labor intensive manual process that lacked inherent learning, and their existing forecasting methods were not tuned across the product life cycle.

A huge portion of workforce hours were spent gathering data and “firefighting” unexpected issues, leaving few resources for increasing accuracy and automation.

With the installation of a Demand Planning Control Tower system, the results speak for themselves:

Coverage of issues increased from 40% to 100%
Accuracy of forecasting increased from 65% to 85%
Frequency of data incorporation went from monthly to a real-time system
Granularity increased from 50% to 100%, with dynamic exploration of best forecasting grain and automatic disaggregation leveraging configurable methods built in
Automation of the process went from 0% to 70%, with automation of the data science work and thresholds to determine cases for human-in-the-loop planning

Refocusing People

Späth concluded that touchless forecasting enables workers to stop generating numbers and gathering data, and instead help build the self-driving supply chain. It’s a mindset shift – instead of people doing work supported by machines, it’s machines doing work guided by people.

This means businesses can take the workforce power previously spent on arduous data-collecting tasks and use it instead for scenario analysis and development of new automation initiatives. This is what the current Age of Agility and Industry 4.0 require, and cognitive automation is making it possible.

‍