case study
A smarter way to slot warehouse inventory, cutting picking labor by 15%
Our client runs tech implementation services for logistics, supply chain, and portfolio management, and works as an Oracle Cloud WMS partner helping warehouses set up the right systems for inventory and order management. For their large-scale warehouse clients, one problem kept surfacing: too much walking, too much labor cost, all coming down to where products got placed on the shelf.
About
- Client: US-based logistics and supply chain tech company
- Industry: Supply chain & logistics
- Location: US
Scope of work
- Data science
- Combinatorial optimization
- Warehouse analytics
Tech stack
- Python
Various products, varying demands, and tedious slotting planning
Our client's warehouses were handling huge, constantly shifting product volumes, with no reliable way to know the best slot for each one in advance. Picking policies varied warehouse to warehouse, and even within the same warehouse, so there was no single formula for minimizing how far staff had to walk
Matching the right product to the right-sized slot, across thousands of SKUs, took long enough by hand that it delayed loading and picking both
Real constraints like cold storage, fragile items, and products that couldn't sit together added more rules than a manual layout could realistically track
Without visibility into tomorrow's likely orders, there was no way to place products where pickers would actually need them next
It works out where each product should be placed in the warehouse, ideally before it's even received, based on where picking staff are most likely to need it.
Because the number of possible product-to-slot arrangements grows so fast that it's a textbook NP-hard problem. Instead of relying only on fixed classical rules, the engine uses combinatorial optimization techniques inspired by AI to search that space efficiently.
The warehouse's physical layout, individual slot sizes, which slots are currently empty, that specific warehouse's picking policy, and historical or forecasted order data.
Both. It also flags "profitable re-slots", cases where moving an already-stored product closer to the staging area is worth the effort.
In a small warehouse, a slightly inefficient layout barely registers. In a large one handling thousands of SKUs and constant in and out cycles, the same inefficiency multiplies into real labor hours and real delivery delays, every single day. Getting slotting right doesn't just save individual picks, it compounds across every order the warehouse processes.
Dynamic arrangement based on future demand and special constraints
STEP
01
Mapping the warehouse
Warehouse geography, slot dimensions, and current available inventory were fed in as the starting map for the model.
02
Learning the order patterns
Historical order data was used to forecast likely near-term demand and surface which products tend to get ordered together.
03
Generating slotting configurations
The engine produced configurations that balanced predicted demand against real constraints, like cold storage needs or products that couldn't be paired, and against what was already sitting in inventory.
RESULTS
15%
Reduction in physical labor for the picking process
The bottom line
Modern warehouses already generate the data needed to solve problems like this, they just haven't had a way to act on it. This slotting engine turns that data into a placement strategy that keeps up with changing inventory and order patterns, cutting down how much ground picking staff have to cover without asking warehouses to change how they operate.
Featured projects
View all
CPG
A mobile-first rebuild with real-time pricing, multi-state compliance, and automation-ready foundations.
Energy10 days mobile MVP shipped
40% mobile traffic served
0 manual pricing uploads
King's College Hospital
Replaced King's College Hospital's legacy EHR registration workflow with a lightweight, modular web platform, improving patient registration, reducing costs, and enabling future-ready compliance.
HealthcareFHIR R4 React frontend on FHIR R4 APIs
2 clicks real-time validation and two-click scheduling
Any device runs on desktop, tablet, mobile, and browser
Clinical Minds
A scalable Selenium + Cucumber framework, integrated into CI/CD.
Healthcare1,000+ test cases created
80% E2E UI automation
75% test cases automated