Can AI-Powered Warehouse Slotting Optimisation Cut Pick-and-Pack Times for Logistics Operators?

AI-driven warehouse slotting optimisation can help logistics operators cut pick-and-pack times by 20-35% — dynamically repositioning inventory based on demand patterns, order profiles, and picker movement data.

The business challenge

In any fulfilment warehouse, where products sit on shelves directly determines how fast they can be picked, packed, and shipped. This placement — known as slotting — is traditionally managed through static rules: fast movers go near packing stations, heavy items go on lower shelves, similar SKUs stay together.

The problem is that demand patterns shift constantly. Seasonal spikes, promotional campaigns, new product launches, and changing order mixes all alter which items should be closest to pick faces. A global logistics operator running five distribution centres across the UK and mainland Europe found that their static slotting rules, updated quarterly by warehouse managers, were consistently 3-4 weeks behind actual demand. Pickers were walking 30% more distance than optimal layouts would require, and pick-and-pack times had crept up by 18% over two years despite stable order volumes.

Why now

Three trends are converging to make AI-driven slotting practical. First, e-commerce order profiles have become more complex — smaller basket sizes, wider SKU ranges, and tighter delivery windows mean picking efficiency directly impacts profitability. Second, warehouse management systems now capture granular data on every pick: timestamps, locations, picker routes, and dwell times. Third, reinforcement learning and combinatorial optimisation algorithms have reached a level where they can solve the slotting problem — which is computationally hard due to the number of possible SKU-to-slot assignments — at practical warehouse scale.

Labour availability adds urgency. With warehouse staff turnover rates running at 30-50% annually across the sector, reducing the walking distance and cognitive load for each pick makes new starters productive faster and experienced pickers more efficient.

The approach

An AI-driven slotting optimisation system typically involves:

1. Data integration — Pull order history, SKU master data, current slot assignments, warehouse layout geometry, and picker movement logs from the WMS and any IoT sensors such as wearable scanners with location tracking.

1. Demand pattern modelling — Build time-series models that forecast SKU-level demand at daily and weekly granularity. The model accounts for seasonality, promotions, day-of-week effects, and trend shifts.

1. Affinity analysis — Identify which SKUs are frequently ordered together in the same pick wave. Placing co-picked items in adjacent slots reduces travel distance per order.

1. Optimisation engine — Use a hybrid approach combining constraint-based optimisation (for hard rules like weight limits, hazmat segregation, and temperature zones) with reinforcement learning (for soft objectives like minimising total travel distance and balancing workload across aisles).

1. Reslotting scheduling — Generate reslotting plans that execute during low-activity windows — overnight or between shifts. The system prioritises moves with the highest impact-to-effort ratio to minimise disruption.

1. Continuous learning — As new orders flow through, the model updates demand forecasts and recalculates optimal positions. Reslotting recommendations become a weekly operational rhythm rather than a quarterly project.

Illustrative outcomes

A transformation like this typically targets:

• A 20-35% reduction in average pick-and-pack time per order
• A 25-40% reduction in picker travel distance per shift
• A 15-20% improvement in new-starter productivity within their first two weeks
• A measurable reduction in picking errors, as logical product grouping reduces mis-picks

What good looks like

• Start with one warehouse: Prove the model on a single site before rolling across the network. Slotting rules vary by warehouse layout, product mix, and operational rhythm.
• Respect operator knowledge: Warehouse managers know things the data doesn't capture — fragile items, awkward packaging, vendor delivery schedules. Build their constraints into the model.
• Measure end-to-end: Faster picking that creates a bottleneck at packing stations doesn't help. Track order-to-dispatch time, not just pick speed.
• Plan reslotting labour: Moving stock takes time. The optimisation engine must account for reslotting cost, not just the theoretical benefit of the new layout.
• Watch for overfitting: A model that over-rotates to last week's demand spike will generate excessive reslotting. Smoothing and minimum-dwell-time constraints prevent churn.

Where Skillikz fits

Skillikz's data & AI team builds optimisation engines that integrate with existing warehouse management systems. We've delivered demand forecasting and process mining solutions for fulfilment operations, and our engineering approach prioritises models that warehouse teams can trust and act on. If your pick-and-pack times are climbing and your slotting rules haven't kept pace with demand, our supply chain risk work shows how we approach these problems. Start a conversation with us.