AI Procurement Benchmarking
Prestige Purchasing
Project Overview
Prestige Purchasing developed an innovative AI-driven platform to help large supply chain companies analyze their procurement data. The software scans messy, unstructured supplier invoices, categorizes the line items, and compares the prices against real-time industry benchmarks.
Before this tool existed, procurement analysts spent weeks doing this work manually in Excel spreadsheets. The ultimate goal of the project—and the QA challenge—was to prove that the AI could extract the data accurately and categorize it correctly much faster than a human analyst could.
Testing Strategy & Focus
I led the testing strategy for the core AI data extraction and categorization pipelines.
• I fed thousands of varied, poorly formatted, and handwritten invoices into the OCR engine to establish a baseline extraction accuracy rate and identify common failure patterns.
• I tested the machine learning categorization logic, ensuring the AI correctly mapped highly specific items (like different commercial brands of flour or cleaning supplies) into the correct standard industry taxonomies.
• I verified the complex price benchmark calculations in the backend, making sure the final frontend analytics dashboards displayed accurate pricing trends and potential cost-saving opportunities.
Business Impact
We successfully proved that the platform's core value proposition worked. Our testing verified that the AI could accurately process a massive procurement workload in minutes that previously took a human analyst two weeks to complete.
By ensuring the underlying data and analytics were mathematically accurate, we gave enterprise procurement teams the necessary confidence to trust the tool and use it for actual, multi-million dollar purchasing decisions.
Need Similar Results?
Let's discuss how we can implement a robust testing strategy for your next product.