CLIENT SUCCESS STUDY: INVENTORY MANAGEMENT
Machine Learning Improves Manufacturing Part Sourcing Process and Saves $900K
AT A GLANCE
Artificial Intelligence improves purchasing process
A global manufacturer struggled with fulfilling the demand for an array of necessary parts due to the impacts of free form text fields in the purchase order process.
By applying rigorous research to product classification, executing process improvements and implementing machine learning to enriched and connected data, the resulting solution matched requested parts with the master catalog and known supplier data. The manufacturer now has access to actionable intelligence and spend analytics for improved efficiencies and savings, with the machine learning and business logic solution improving over time for long-term value.
savings due to ML and actionable intelligence.
accuracy after two rounds of training.
efficiency gains for FTEs using the ML model and tools.
Download full report
The Need for Change
Free form text fields used to request parts unlisted in a manufacturer’s material master made tracking, fulfilling and anticipating demand for these products difficult, and impacted reporting accuracy and ultimately the ability to negotiate prices. The manufacturer was unsure of the scale of the issue or how to tackle it within their system.
They wanted to understand:
The expanse and effects of FFT across the business
The detailed gaps in procurement for category managers
The comprehensive impact on spend
Understand the cost of errors and price differences caused by free form text inputs.
Capture customer demand by ensuring product pricing and availability.
Optimize savings and negotiate best rates from suppliers.
Apply Machine Learning for intelligent automation.
Automating smart learning for faster categorization and ordering
Aligned Automation applied deep product research and extracted detailed classification, working with the manufacturer to categorize products and develop business logic. An AI/ML algorithm library was inserted into the material master to extract part numbers, check for similarities and ultimately recommend the best matches. This led to a Part Extraction Engine with the ability to continuously learn and improvise.
Utilizing Natural Language Processing, the Part Extraction Engine “read” short and long free form text descriptions, matching requested parts with those in the material master as well as with connected supplier data. With this machine learning-driven triangulation, automatically detecting and ordering the correct part at the best possible price could be achieved.
Optimized long-term savings and efficiencies
In 90% of the cases the ML engine was able to correctly identify the part from FFT, verify stock and pricing with suppliers, add the item to the material master and order. Manual effort was cut by 50%, with constant improvement to speed and quality.
The team used the data to create a monthly reporting dashboard to track FFT requests and spend. The manufacturer can register spending and ensure category managers are able to control the full volume of customer demand, while freeing employees for higher-value work.
With accurate spend reporting and actionable intelligence, the company can better predict customer needs and negotiate consistently with OEMs and vendors. Over time, the machine learning and business logic solution will become faster and more effective at accounting for FFT field, leading to long-term value savings.