Amazon: Optimizing Non-Sortable Fulfillment Centers with AI-Driven Initiatives
Led analytics and optimization for a $7.1M waste reduction project in non-sortable fulfillment centers. The 'Stick' initiative alone drove $3.1M in savings through combined stow and pick processes.
Connect on LinkedIn
Project Overview
As Product Manager for AI at Amazon in 2019, I spearheaded key contributions to the TNS Optimization project. My work included data reporting, process improvements, and rollout recommendations.
This case study details how we addressed inefficiencies in aisle changes, buffer travel, cage changeovers, and bin searches. The result? Multimillion-dollar impact and enhanced operational efficiency across European fulfillment centers.
This showcases my expertise in scaling AI-driven solutions for enterprise logistics and complex operational challenges.
$7.1M
Total Savings
Annual waste reduction achieved
$3.1M
Stick Initiative
Core program impact
3
Fulfillment Centers
DTM1, MXP3, LBA2
The Problem: Inefficiencies in Non-Sortable Fulfillment Operations
In 2019, Amazon's non-sortable fulfillment centers faced $7.1M in annual waste. Sources included long aisle changes, buffer travel time, frequent cage changeovers, and manual bin searches.
Low Staffing Compliance
Only 64-81% compliance due to managers' underuse of labor data and inadequate workforce allocation tools.
Low Process Compliance
Just 41-80% adherence from associates' errors or non-adherence to standard operating procedures.
Root Causes
Inadequate onboarding, unclear instructions, and lack of accountability systems risked sustained inefficiencies.
Rollout Risk
These challenges threatened expansion across European FCs like DTM1, MXP3, and LBA2.
Streamlining Operations: Analytics-Driven Optimization
I collaborated with cross-functional teams to analyze pilot data, resolve ETL issues, and refine processes. Our focus was the 'Stick' initiative—combined stow-pick workflows to minimize waste.
My contributions included maintaining daily reports for decision-making across three FCs. I modified queries for hourly visibility to enable shift-level accountability.
I participated in stakeholder calls to align on priorities and ensure data-driven roadmaps. This set the foundation for sustainable operational excellence.
01
Data Infrastructure
Resolved ETL issues and established reliable reporting systems
02
Hourly Tracking
Modified queries for shift-specific insights and accountability
03
Training Enhancement
Developed dashboards and visual aids for associate clarity
04
Compliance Coaching
Added targeted coaching to achieve 100% adherence
Key Innovations: AI-Enhanced Workflows and Compliance Tools
Combined Stow and Pick
Integrated stow-pick in the same cage, boosting units per changeover by 69.5%. Stow UPH increased 20.2%, pick UPH rose 14.0% in LBA2 tests.
Hourly Data Reporting
Modified queries to provide shift-specific insights. Enabled precise identification of compliance issues and targeted follow-ups.
Compliance Dashboard
Universal tool for FCs to track key metrics. Monitored stick percentage and least compliant associates, leads, and managers.
Process Visual Aids
Stickers in PUP cages and instructional videos reduced information loss. Ensured consistent associate training across shifts.
"These optimizations transformed wasteful processes into efficient, data-driven systems, proving AI's role in scaling fulfillment operations."
Results: Multimillion-Dollar Savings and Operational Gains
69.5%
Units Per Changeover
Increase in efficiency
20.2%
Stow UPH
Productivity uplift
14.0%
Pick UPH
Performance gain
The TNS Optimization project delivered $7.1M total waste reduction. The 'Stick' initiative alone contributed $3.1M in savings.
In LBA2 pilot, 8.9% stick-eligible volume was processed. This led to dramatic improvements in units per changeover and UPH metrics.
Hourly reports enabled accountability, reducing non-compliance from 40-80% range. Quality remained stable with no significant DPMO impact.
Buffer Optimization
$1.1M saved through improved buffer management and reduced travel time
Guided Stow
$1.7M saved via AI-directed stow processes and bin optimization
Sort at Receive
$1.3M saved by streamlining receiving and sorting workflows
Key Takeaways: Building Resilient AI Optimization in Logistics
1
Data ownership drives compliance
Empowering FC teams with accessible tools like dashboards ensures sustained process adherence. When teams own their data, accountability increases naturally.
2
Iterative training is key
Clear onboarding and visual aids minimize errors across diverse workforces. Coaching reinforces 100% compliance and builds lasting habits.
3
Balance metrics with root causes
Addressing associate behaviors and managerial tools unlocks full entitlement. Surface-level metrics aren't enough without understanding why.
4
Cross-functional alignment accelerates impact
Stakeholder discussions and ETL resolutions enabled rapid problem-solving. Breaking down silos creates momentum and drives results faster.
"This project at Amazon refined my approach to AI PM, emphasizing data-driven roadmaps that deliver scalable, high-ROI solutions in complex environments."
Looking Ahead: Scaling AI Optimization in Enterprise Systems
The TNS Optimization experience informs my work in AI product management today. I focus on proactive workflows that reduce waste and enhance efficiency.
I'm applying these insights to projects like PipelineOM, a sales AI operating system. I'm actively seeking roles at leading AI companies to orchestrate similar transformations.
Open to collaborations on optimizing enterprise operations and scaling AI-driven logistics solutions. Let's connect to explore opportunities.
1
Early 2019
Project Initiation & Pilot Planning
2
Mid-2019
Data Analysis & Reporting Setup
3
Late 2019
Pilot Execution & Compliance Improvements
4
Post-2019
Rollout Recommendations & Sustained Impact
Ready to discuss AI optimization and enterprise logistics?