How Shipsy's WMS slots inventory by velocity, affinity, and pick-path physics

Slotting is the quiet superpower of a high-throughput warehouse. Shipsy’s WMS reslots continuously — not quarterly — using velocity, SKU affinity, weight/ergonomics, and pick-path travel physics to place every SKU where a picker travels least to get to it. On real enterprise implementations this cuts pick travel by 20-40%, the single biggest lever in warehouse labor cost.

Why we built this

Slotting decisions made at warehouse go-live are usually optimal for week one and slowly decay for the next three years. Demand changes — new SKUs, seasonal shifts, changing bundle patterns — but most WMS implementations reslot manually at best, quarterly, and even then only on the obvious fast movers. The cost is invisible in the P&L but massive: pickers walking two or three times the necessary distance, shift after shift.

A leading 24/7 omnichannel pharmacy chain in Indian metros, a major Middle East 3PL and contract logistics provider, and a global biotech scaling rare-disease therapeutics across 30+ countries all flagged slotting drift as a top-three throughput constraint. Shipsy’s continuous slotting engine treats slotting as a live optimization, not a setup task.

How it works

Velocity analysis. Every SKU gets a velocity profile computed daily: pick frequency (times per day), pick quantity (units per pick), outbound pattern (time-of-day concentration), and trend (rising, falling, stable). Velocity tiers (A, B, C, plus slow-movers and dead stock) are recomputed continuously.

Affinity clustering. SKUs that co-pick (appear in the same order frequently) get an affinity score. When two SKUs have high affinity — e.g., printer + ink cartridges, frequent bundle SKUs — placing them in adjacent slots cuts pick travel significantly. Shipsy computes pairwise affinity across the full SKU catalog and surfaces the top co-pick pairs for slot placement.

Pick-path physics. The slotting engine models the warehouse as a graph — aisles, racks, bin locations, travel distances, congestion zones. When considering a slot move, it doesn’t just check “is this closer to the pack station”; it simulates actual pick-path traversal for the next week’s projected pick list and scores each candidate placement on total travel time.

Weight and ergonomic rules. Heavy SKUs place at ground-level or pick-cart-height — not because of a simple rule, but because the scoring function heavily penalizes overhead-heavy or under-knee-heavy placements. Fragile items go to stable slots; hazardous items go to segregated zones with access rules. The engine treats these as hard constraints rather than soft preferences.

Temperature and compliance zones. For regulated warehouses (pharma GDP, cold-chain FMCG), SKU placement respects temperature-zone assignment and 21 CFR / GDP audit requirements. A global pharma CDMO uses Shipsy’s slotting engine with compliance-zone overlays to maintain GDP audit posture without human reslotting effort.

Continuous reslotting, not big bang. Instead of a quarterly “reslot everything” event, the engine surfaces slot-move recommendations every morning — typically 20-60 high-value moves that recover the most picker-travel. Warehouse ops can accept, reject, or batch them into the next replenishment cycle. Small, continuous moves avoid the operational disruption of a big-bang reslot.

Wave planning integration. Slotting feeds directly into wave planning — see wave planning + pick-path optimization. When waves are generated, the pick-path optimizer uses the current slotting layout. As slotting improves continuously, wave efficiency compounds.

Dead-stock detection. SKUs that haven’t moved in a configurable window (30/60/90 days) are auto-flagged as dead stock with recommendations: move to low-cost storage, move to remote bin, liquidate, or return to supplier. Dead stock in prime locations is a silent margin killer; continuous detection makes it visible.

New-SKU slotting. When a new SKU arrives (first ASN for a new product), the engine predicts velocity based on similar SKUs and assigns an initial slot aligned to the prediction. If actual velocity diverges from prediction in the first weeks, reslot recommendations fire automatically.

Early results

Warehouses moving from periodic manual slotting to continuous engine-driven slotting typically see: pick travel reduced by 20-40%, picks-per-labor-hour up 15-30%, and dead stock in prime slots down to <5% of prime capacity (from industry-typical 15-25%). A major Middle East 3PL and contract logistics provider uses this pattern across multiple client tenants in a multi-tenant warehouse, where slotting drift compounds across tenants. See related post on 3PL multi-tenant operations.

For e-commerce fulfillment centers running high SKU counts (50,000+ SKUs), the affinity-clustering effect is the biggest single contributor — co-picking SKUs placed adjacent saves cumulative miles per shift.

What’s next

Next release integrates demand-forecast signals: slotting decisions weight not just last-30-days velocity but next-30-days projected velocity, so fast-risers get prime placement before the wave hits. For retailers with known promo cycles, this cuts peak-season reslotting scramble to near zero.