Why Picking Strategy Is a Make-or-Break Decision
Order picking is the most labor-intensive activity in any warehouse. It accounts for 50–65% of total warehouse operating costs and consumes more labor hours than receiving, put-away, replenishment, and shipping combined. ↗
A typical picker in a conventional warehouse walks 15–20 km per eight-hour shift, yet spends only 15–20% of that time actually picking. The rest is travel, searching, and confirmation tasks. ↗
Choosing the right picking strategy, and sequencing it correctly, is the single fastest lever for cutting labor cost and improving throughput without touching automation hardware. This guide covers every major strategy, when to use each, how to combine them, and how to implement improvements in a realistic 30-day window.
Before evaluating specific methods, it helps to understand what drives picking cost. Three variables dominate:
- Travel distance is the largest controllable cost. Research consistently shows that 30–50% of a picker's time is spent walking between pick locations. Route and batching optimization directly attacks this variable.
- Order concurrency — how many orders a picker handles per trip — determines equipment needs and sorting complexity. Discrete picking (one order per trip) is simple but expensive. Batch or cluster picking multiplies throughput per trip but requires effective sorting at pack.
- Labor flexibility is the hidden cost that most strategies ignore. A strategy that works well with 20 trained pickers may break down with 40 seasonal temps. Methods that depend on deep warehouse familiarity don't survive peak periods.
Getting strategy right compresses cost on all three dimensions simultaneously.
The 6 Core Picking Methods Explained
1. Discrete Picking (Single-Order)
The simplest method: one picker, one order, one trip.
The picker works through a single order from start to finish, then returns to the pack station before starting the next.
Best for: Small warehouses (fewer than 5,000 orders per day), low SKU count, or operations with highly variable order profiles where batching doesn't produce clean groups. Also appropriate when order accuracy is paramount and sorting errors at pack would be costly.
Limitations: Highly inefficient at scale. If the same picker walks to the same aisle five times for five different orders, four of those trips are waste.
2. Batch Picking
A picker collects items for multiple orders in a single warehouse trip, typically using a multi-compartment cart or tote system.
After the trip, items are sorted by order at the pack station. Research by Gademann, van den Berg & van der Hoff (IIE Transactions, 2001) found that well-structuredbatching reduces travel distance by 27–40% compared to discrete picking on the same order set.
Ideal batch size: Academic literature and operational benchmarks converge on 8–15 items across 3–8 orders as the efficiency sweet spot for manual batch picking. Beyond roughly 15 orders per batch, the sorting burden at pack typically erodes the travel savings.
Best for: High-volume e-commerce and B2C fulfillment with many small orders sharing common SKUs. Works best when fast-moving SKUs are concentrated in a relatively small area.
Limitations: Requires good sorting discipline. Mis-sorts at pack directly create order errors. Batching also requires a mechanism for grouping orders intelligently — random batching produces minimal travel savings.
3. Cluster Picking
A variant of batch picking where picks are grouped by location rather than by time or order composition.
The picker carries a multi-tote cart (often 6–12 totes) and fills each tote for a different order simultaneously as they move through the warehouse. Items drop directly into the correct order tote at the pick face - no sorting required at pack.
Key distinction from batch picking: Batch picking sorts after the trip. Cluster picking sorts at the pick face during the trip. This eliminates the pack-station sorting bottleneck but demands enough order density in each zone to justify the multi-tote cart configuration.
Best for: Operations with medium-velocity SKU profiles, where orders frequently share aisle locations. Works particularly well for B2B fulfillment with predictable order patterns.
Limitations: Cart configuration matters enormously. Too few totes and you don't capture enough concurrency; too many and the cart becomes unwieldy, especially in narrow aisles. Cluster picking also performs less well when items are large or irregularly sized.
4. Zone Picking
The warehouse is divided into logical zones, with pickers assigned exclusively to their zone.
Orders move between zones - either sequentially (zone-to-zone conveyor or tote handoff) or in parallel (wave-synchronized zone picking with later consolidation). Each picker becomes a specialist in their zone's product assortment.
- Sequential zone picking (pick-and-pass): The order document or tote travels from zone 1 to zone 2 to zone 3. Simple to manage, but creates bottlenecks if any zone becomes a constraint.
- Parallel zone picking: All zones pick simultaneously. Completed zone picks are consolidated downstream. Faster throughput, but requires a merge/consolidation step before packing.
Zone picking's primary advantage is specialization. Pickers who work the same area daily learn product locations, reducing search time and error rates. ↗
Best for: Large warehouses with 30,000+ SKUs, operations with temperature or handling requirements that require physical separation, and any operation with strong seasonal velocity variation by category.
Limitations: Creates interdependencies between zones. One slow zone delays the entire order. Zone boundaries require periodic rebalancing as velocity profiles shift.
5. Wave Picking
Orders are grouped and released to the floor in scheduled waves — typically 2–6 waves per shift — aligned with shipping schedules, carrier cutoffs, or labor availability.
Within each wave, pickers are often assigned zones or product categories. Wave picking is less a standalone picking method and more an orchestration layer that sits on top of batch, zone, or discrete picking.
A 2024 simulation study in Scientific Reports (Nature) found that consolidating orders into waves of optimized sizes achieved up to a fourfold reduction in total pick distance for the studied dataset — though the authors noted that effectiveness is highly sensitive to wave size and order clustering.
Best for: Operations with multiple daily carrier cutoffs, parcel carriers with hard sortation windows, or any fulfillment operation that needs to align labor deployment with output requirements in real time.
Limitations: Wave picking requires planning discipline. Poorly constructed waves — too large, too small, or misaligned with labor — add administrative overhead without delivering the efficiency benefit. It also increases order cycle time: orders wait for the next wave rather than being picked immediately on receipt.
6. Case Picking (Full-Case and Layer Picking)
Case picking handles full shipping cases or pallet layers rather than individual units.
The picker moves through the warehouse pulling full cases (typically onto a pallet or slip-sheet) rather than picking eaches. Often used in B2B and wholesale distribution where customers order by the case.
Layer picking — pulling full layers from pallet positions — represents the next tier of scale, typically requiring lift equipment and moving into semi-automated territory.
Best for: B2B distributors, food and beverage distribution, wholesale fulfillment, and any operation where a meaningful percentage of order lines are full-case quantities.
Limitations: Does not address each-level picking, which remains the most labor-intensive operation. Many warehouses run case picking and each picking as separate workflows.
Picking Path Strategies
Picking method determines how many orders a picker handles per trip. Picking path strategy determines what route the picker follows through the warehouse. Both matter, and they operate independently — any method can be combined with any path strategy.
S-Shape (Serpentine) Routing
The picker traverses aisles sequentially — entering at one end, exiting at the other, then crossing to the next aisle. This eliminates backtracking within aisles and is the easiest manual routing policy to implement.
When to use: Warehouses with 5+ parallel aisles and reasonably uniform pick density across the floor. Time savings versus random routing: 15–30% in typical implementations.
Limitation: S-shape is suboptimal when picks are concentrated in a subset of aisles, because the policy still forces full traversal of each aisle in the sequence. ↗
Return Routing (U-Shape)
The picker enters an aisle, travels to the deepest pick point, and returns out the same entrance rather than exiting at the far end. This allows aisles to be skipped entirely when they contain no picks.
When to use: When pick density is low or unevenly distributed — typically during off-peak periods or for slow-moving SKU sections. Time savings: 20–35% versus random routing.
Limitation: Requires discipline in identifying aisle skip opportunities. In practice, this is difficult without software computing the optimal skip points per trip.
Largest Gap Routing
For each aisle that contains picks, the picker enters to the pick point nearest the depot-end, traverses to the pick point nearest the far-end, then decides whether to continue through or return — based on where the largest gap (longest stretch without a pick) is in that aisle. The aisle section with the largest gap is excluded from the route.
When to use: Medium-density warehouses with 6–15 aisles. Produces near-optimal routes without requiring full combinatorial optimization. Time savings: 25–40% versus random routing.
Combined Routing
Most real warehouse layouts don't conform to any single routing policy. Combined routing uses S-shape for high-density sections and return routing for low-density tails, with zone boundaries to limit each picker's travel universe.
This is the practical approach for most warehouses above 5,000 sq ft with heterogeneous velocity profiles.
A real implementation example: a 5,000 sq ft e-commerce fulfillment center that switched from random picking to S-shape routing combined with 3-order batching reduced average pick time per order from 12 minutes to 8 minutes — a 33% reduction — with no software, no WMS change, and no capital expenditure.
Head-to-Head: When to Use Each Strategy
Choosing between methods requires matching strategy capabilities to your operation's specific profile. The critical variables are order volume, average lines per order, SKU count, warehouse size, and whether you have a WMS with optimization capability.
Batch size benchmarks: Batch picking delivers maximum efficiency when the coefficient of variation of pick locations across orders in the batch is low — meaning orders in the same batch tend to need items from the same areas. Random batching produces only 10–15% travel savings; intelligent batching (grouping orders whose pick locations are geographically close) consistently produces 30–45%.
How to Choose the Right Strategy for Your Warehouse
No single method dominates all scenarios. The decision framework below covers the four most common decision points:
- Step 1: Measure your baseline travel fraction.
Time-study 10 pick trips. What percentage of time is travel versus actual picking? If travel exceeds 40%, any batching or routing improvement will produce immediate, measurable gains. If travel is already below 25%, the bottleneck is likely elsewhere (put-away accuracy, pick face density, confirmation workflows).
- Step 2: Analyze your order profile.
What is your median lines per order? If most orders have 1–3 lines, discrete picking with good routing may be sufficient. If orders average 8+ lines, cluster or zone picking become attractive. What percentage of orders share at least one SKU? High SKU overlap between concurrent orders is the primary enabler of efficient batching.
- Step 3: Evaluate your SKU velocity distribution.
In most warehouses, the top 20% of SKUs by velocity account for 80% of picks (the Pareto concentration pattern). If your velocity distribution is highly concentrated, slotting your fast-movers close to the pack station may deliver more savings than any picking method change. If velocity is evenly distributed across the warehouse, routing optimization is more important than slotting.
- Step 4: Consider labor profile
High-turnover environments with frequent temp workers favor simpler methods (discrete or cluster) that can be learned in under an hour. Operations with stable, experienced picker pools can exploit the efficiency gains of wave + zone picking, which requires more training and floor coordination.
Adding Route and Batch Optimization Software
Manual routing policies and visual batching rules deliver meaningful gains but have a ceiling.
They can't account for real-time inventory availability, mid-wave order insertions, aisle congestion, or the compound effect of optimizing batching and routing simultaneously.
Software-driven optimization removes this ceiling.
The gains compound: route optimization alone reduces walk distance by 20–30%.
Intelligent batching alone reduces it by 15–25%.
Combined optimization — where the software selects batches and sequences routes across the full pick floor simultaneously — typically achieves 30–55% reduction, because it avoids local optima that arise when the two problems are solved independently.
Walk distance per picker per shift drops by an average of 1.0–1.5 km.
Orders per picker per hour increase from a typical 12–18 to 22–35, depending on baseline conditions.
Pick error rates fall because optimized routes reduce the per-order complexity a picker manages.
Onboarding time for new pickers drops from 2–3 weeks (learning the warehouse) to 1–3 days (following system-generated instructions). At 5,000 orders per day, the reduction in required picker headcount is typically 20–35%.
Optioryx Pulse is an AI-powered warehouse optimization platform that applies this combined approach - simultaneous route optimization, intelligent order batching, and slotting recommendations - without requiring WMS replacement.
It integrates via API with existing WMS systems and delivers optimized pick sequences directly to pickers' existing devices or scanners.
When software optimization pays off
For operations processing 50,000+ picks per day, the ROI calculation is typically straightforward.
At that volume, a 20% improvement in picks-per-hour translates to roughly 2–4 fewer FTEs per shift, with annual labor savings well in excess of software cost.
Below 10,000 picks per day, the business case depends more on error rate costs, temp worker dependency during peaks, and the value of the onboarding time reduction.
Questions?
No single method is most efficient for all warehouses. For high-volume e-commerce operations (5,000–25,000 orders/day), a combination of zone picking with intelligent batch picking within zones, supported by route optimization software, consistently produces the best results. For smaller operations (under 5,000 orders/day) with concentrated SKU velocity, discrete or small-batch picking with S-shape routing often delivers the best balance of efficiency and simplicity. The key variables are order volume, average lines per order, SKU velocity distribution, and your labor profile.
Zone picking divides the warehouse into fixed areas with a dedicated picker assigned to each zone. Orders either travel sequentially from zone to zone (pick-and-pass) or are picked in parallel across all zones and consolidated downstream. Zone picking works best for large warehouses with 30,000+ SKUs, operations with temperature or handling requirements that require physical separation, and fulfillment centers where picker specialization by product category reduces search time and error rates. The trade-off is zone balance — one slow zone can delay the entire order.
Both involve collecting items for multiple orders in a single warehouse trip. The difference is where sorting happens. Batch picking gathers items into a single container and sorts them by order at the pack station after the trip. Cluster picking uses a multi-tote cart — typically 6–12 totes — and drops each item directly into the correct order tote at the pick face during the trip. Cluster picking eliminates the pack-station sorting step but requires higher pick density per trip to be efficient. Batch picking is simpler to set up; cluster picking is faster at scale when order density supports it.
Academic research and operational benchmarks consistently show 20–55% reductions in picker travel distance when route optimization is applied. Manual routing improvements — switching from random to S-shape or return routing — typically deliver 15–35%. Software-driven combined optimization (simultaneously optimizing routes and order batching) delivers 30–55%. The higher end of the range requires both route and batch optimization running together, because optimizing them independently produces local optima that combined optimization avoids.
Yes. Manual routing improvements (S-shape traversal, return routing for low-density aisles, visual zone markers) and manual batching rules (grouping orders by pick area before releasing to the floor) can be implemented with paper pick lists and basic order management systems. A warehouse that switches from random picking to structured S-shape routing combined with simple 3-order batching typically sees 20–35% reductions in pick time and travel distance within weeks, with no software and no capital expenditure. Software optimization raises the ceiling to 30–55% reduction but is not required for meaningful first-step improvements.