Client’s extensive product portfolio, combined with its international distribution, created massive product attribution challenges and equally massive datasets. Consequently, the client had been spending anywhere from 32 to 128 labor-hours per month assigning and/or updating these fields. In addition to straining resources and extended lead-times, the accuracy-rate (across 13 countries) averaged 80%.


Atlas software engineers worked with the client’s brand management personnel to automate the product attribution process for every SKU the client sold in thirteen countries. Collectively, the team developed an Auto-Attribution solution that used rules-based routines and machine learning to minimize product attribution costs and lead-times, while improving accuracy.


Product attribution accuracy increased from 80% to 99% with continuous improvement functionality built into the solution’s machine learning capability. More importantly, the number of labor-hours required for managing product attribution data for the client’s extensive product line was reduced from 80 hours (average) to one hour — an efficiency gain of 98.8%.