Inside Shipsy's wave planning: how picks and packs get batched for maximum throughput

Wave planning is how modern warehouses turn thousands of orders into coordinated picker movement. Shipsy’s wave engine generates pick waves that respect order cutoffs, pick-path physics, pack-station load balance, and vehicle dispatch times — all at once. The output is a measurably faster order-to-dock cycle with less idle labor and fewer late shipments.

Why we built this

The default wave-planning approach in legacy WMS is “release all orders every 30 minutes and let pickers figure it out.” The cost shows up as: pickers criss-crossing aisles, pack stations backed up while pickers stand idle, orders missing outbound truck cutoffs because the wave didn’t sequence them correctly, and overtime shifts to absorb the chaos on peak days.

A global big-and-bulky retailer leading in furniture and home goods, a leading 24/7 omnichannel pharmacy chain in Indian metros, and a global 3PL with European roots all flagged wave-planning quality as a primary throughput constraint. Shipsy’s wave engine treats wave design as a constrained optimization problem, not a fixed-interval dump.

How it works

Order pool intake. Every open order flows into the wave-planning pool with attributes: SKU list, quantities, pack type (single-SKU, multi-SKU, kit), outbound carrier/service, cutoff time, priority tier, and destination cluster. The pool is continuously updated as new orders land and existing orders fulfill.

Wave trigger logic. Waves fire based on multiple triggers: outbound cutoff approaching (e.g., 2 hours before the 6pm courier pickup), pack-station utilization below threshold (triggering a pre-emptive wave to keep packers busy), labor-shift boundary, or manual override for exception cases. Cutoff-driven waves include only orders that must ship on the current cutoff; fill-driven waves pick orders that maximize pack-station throughput.

Pick-path optimization inside the wave. Once a wave is sized, the engine batches orders into pick tasks. The core decision: which orders go to which picker, in what sequence, and along what pick path. The optimizer minimizes cumulative pick travel across the wave — not per-picker, but across the entire wave’s work bundle.

Picks are batched three ways: discrete (one order per picker per trip), batch-pick (multiple orders picked simultaneously, sorted at pack), and zone-pick (picker stays in a zone, orders flow through). The engine picks the strategy per wave based on order composition.

Pack-station load balancing. Output from pickers flows to pack stations. The wave engine pre-allocates pack-station capacity — each station’s queue is modeled so no single station gets slammed while others stand idle. For waves with heavy single-line orders, pack allocation skews to fast-pack stations; for multi-line kit-pack orders, allocation skews to stations with kit-build tooling.

Slotting awareness. Pick-path optimization consumes the current slotting layout (see WMS slotting optimization). As slotting improves continuously, wave efficiency compounds automatically — better slots mean shorter optimal pick paths.

Outbound integration. Each wave is tagged with its outbound truck or courier pickup window. When the wave completes, packed cartons auto-stage at the correct outbound dock. Shipsy’s dispatch module then matches packs to load plans — see vehicle capacity + load planning — and loads the outbound in reverse drop order for multi-stop routes.

Exception handling mid-wave. If a picker hits a stock-out (bin empty when expected to have inventory), the wave engine reassigns the line to an alternate bin or another picker currently in the right zone — without blocking the rest of the wave. This is the “live re-optimization” that legacy wave planners can’t do because they batch-release and walk away.

Labor balance. For each wave, the engine allocates labor: number of pickers per zone, number of packers per station, and one or two flex-labor assignments for recovery. Allocation respects breaks, shift boundaries, and ergonomic constraints (no picker stays on heavy-lift zones for their full shift).

KPI telemetry. Every wave emits telemetry: wave size, pick-travel total, average pick rate, pack-station utilization, on-time completion. Telemetry feeds Atlas, Shipsy’s control tower, so anomalies (a wave running 30% slower than forecast) raise operational flags in real time. See real-time incident management.

Early results

Warehouses moving from fixed-interval wave release to constrained wave optimization typically see: picks-per-labor-hour up 20-35%, on-time outbound departure rate up 10-20 percentage points, pack-station utilization up from 60-70% to 85-90%, and overtime hours cut 30-50% on peak days. A major Middle East 3PL and contract logistics provider uses this pattern across multi-tenant warehouse operations where tenant-level order cutoffs vary by service contract.

For e-commerce operators with same-day cutoffs, the cutoff-driven wave logic directly prevents missed-cutoff incidents — the single biggest cause of customer-facing SLA breaches in D2C fulfillment.

What’s next

Next release introduces predictive wave modeling — the engine simulates the next 4 hours of waves continuously and surfaces “if you reallocate labor to zone 3 now, you’ll save 40 pick-minutes in the 4pm wave” recommendations to the floor supervisor.