Summary
Warehouse picking systems are how a warehouse gets items from storage into orders. Every system is three choices: the method, the equipment and who makes the daily decisions. The equipment either sends pickers to the stock (picker-to-parts) or brings the stock to them (parts-to-picker). Choose it by your order profile, then check who decides batches, routes and slots, because that is where many sites lose time without noticing.
Most picking is still done by people who walk. Only 12% of operations report fully automated picking, according to the 2026 Peerless Research Group study for Supply Chain 24/7. A third say their picking is mostly or fully manual with no plans to change that.
So when a site outgrows its setup, the useful question is which mix of equipment fits the orders you actually ship. This guide compares six types of warehouse picking systems, gives a decision matrix to narrow them down and shows how to test the options on your own orders first.
What is a warehouse picking system?
A warehouse picking system combines how you group orders, the equipment pickers use and the logic that plans the work. The three are separate choices. A site can move from single-order picking to cluster picking without buying anything, or add pick-to-light without changing how orders are grouped.
The method is how orders are grouped: one order per tour, several orders on one cart, zones that pass totes along, or waves timed to carrier cut-offs. Our guide to picking methods and how to choose one covers these, so this article stays with the other two.
The equipment is how the picker and the stock meet. In picker-to-parts systems, the picker walks or drives to the location. In parts-to-picker systems, also called goods-to-person, a machine brings a bin to a fixed station.
The decisions are the daily planning the equipment does not do on its own: which orders share a tour, the route, which items sit in the fast spots and when work hits the floor. Picking can take up to 55% of a warehouse's operating costs, according to a 2023 study in the journal Logistics, so these decisions show up directly in labor hours.
Most sites run a mix. In the 2025 edition of the Peerless survey, 42% of companies used a mix of automated and manual processes for order fulfillment and 52% were still mostly or all manual, as Logistics Management reported.
Types of order picking systems
Order picking systems fall into two families, with six common types of warehouse picking technology between them. Picker-to-parts makes the walk shorter or the pick faster; parts-to-picker takes the walk away, at the cost of fixed capacity and a large investment.
Picker-to-parts: the picker walks
A cart or trolley with paper or an RF scanner runs on almost any layout, item size and order mix, and it is the cheapest setup to start and to change. How far pickers walk depends almost entirely on the batches and routes they are given.
With voice picking, a headset reads out location and quantity, which keeps hands and eyes free for cases, cold stores and heavy items. In 2025, 21% of companies in the Peerless survey used it. Voice changes how instructions reach the picker, not how far they walk.
With pick-to-light, a light at each location shows what to pick, and put-to-light walls do the same for sorting a batch into orders. Every location needs hardware, so it suits a fixed fast-moving zone better than a large or changing range. Light-based tools were in use at 42% of respondents in the 2026 study.
Collaborative AMRs carry the totes between pickers, who stay in their own zone. The robots remove the walking with a full cart and the trips to the drop-off point, but they need clear aisles, and a fleet sized for peak sits partly idle the rest of the year.
Parts-to-picker: the stock comes to a station
Goods-to-person storage covers cube storage grids, shuttles, vertical lift modules and carousels. These automated warehouse picking systems store densely and remove walking for the items inside them. Capacity is fixed by the ports, bins and robots you buy, lead times run into months, and bulky items stay outside.
In robotic piece picking, an arm picks from bins at a station, usually fed by goods-to-person storage. It works for items the arm can grip reliably, and people handle the exceptions.
| Type | How it works | Fits best | Watch out for |
|---|---|---|---|
| Picker-to-parts: the picker walks | |||
| Cart with paper or RF scanner | Picker walks with a cart and scans each pick | Any layout and item mix; changing ranges | Walking depends on batches and routes |
| Voice picking | Headset gives location and quantity | Cases, cold stores, heavy items | Same walking as a cart |
| Pick-to-light / put-to-light | Lights show where and how many | Small, fast items in a fixed zone; sorting batches | Hardware at every location |
| Collaborative AMRs | Robots carry totes between pickers | Long walks to packing; many small orders | Clear aisles; fleet sized for peak |
| Parts-to-picker: the stock comes to a station | |||
| Goods-to-person storage | Bins come to a fixed station | Small, fast items; dense storage | Fixed capacity, long lead time, big investment |
| Robotic piece picking | Arm picks from bins at a station | Items an arm grips reliably | Exceptions go to people |
How to choose a warehouse picking system: a decision matrix
The orders you ship decide this more than the size of the building. Six measures from your own order history narrow the options down, and most come straight out of a WMS export:
- Lines per order. Short orders favour batching and goods-to-person; long orders favour walking the aisles.
- How concentrated the picks are. A small group of SKUs that takes most picks is a candidate for lights or automated storage.
- Volume and peaks. Automation is sized for either a normal day or your busiest week, and you pay for the gap.
- Item sizes and weights. Pallets, long items and heavy cases rarely fit bins, ports or grippers.
- Share of the shift spent walking. Under a quarter, and the time goes elsewhere: searching, confirming or packing.
- How often the mix changes. New clients and seasonal swings punish fixed equipment.
In the matrix, a half dot means the fit depends on details worth testing. We left out volume thresholds on purpose: published cut-offs are mostly vendor rules of thumb, and your order data answers the question better.
Where to start, by situation
With many small orders and a lot of walking, start with better batching on the carts you have, then look at AMRs if the walk to packing is still the biggest loss.
When a small group of SKUs takes most of the picks, give that group its own zone with lights or automated storage and keep the long tail on carts. Many mixed sites are split this way.
With a changing range, many clients or bulky items, stay with carts or voice and put the effort into the decisions. That is the core of our piece on why optimizing first protects an automation business case.
Goods-to-person has its strongest case with high, steady volume of small items and few people to hire, as long as the items fit the bins.
What every picking system still leaves open
No equipment decides which orders go together, where items sit or when work goes to the floor. Who makes those decisions changes the result as much as the equipment does. A cluster cart on a poor route can walk further than a single-order picker on a good one.
On carts and voice, the decisions usually come from fixed WMS rules, such as sorting each pick list by a location number set up years ago. That is where most picker-to-parts sites can win time back without new equipment. Our article on how an optimization layer differs from a WMS and a WES explains who does what.
Mixed sites need two answers: the automated zone runs on its own software, and the rest of the building runs on whatever plans the manual picking.
Buyers ask about the equipment and the planning at the same time. A Fortune 500 company automating its largest US distribution center said on a call with us: "Currently there is a mix." The same team asked whether software could "suggest which type of equipment should be used and modify the amount of equipment". On Warehouse Wizards we talked about people and machines working side by side in sites like these.
Test the options on your own orders before you buy
Each vendor of warehouse picking solutions models its own system. Replaying your own orders on a model of your layout puts every option on the same data, and shows walking, picks per hour and labor hours before anything changes on the floor. All it takes is an export of order history and a layout drawing, with no integration.
- Export two to four weeks of picked orders, busy days included, and draw the layout with its aisles, one-way rules and drop-off points.
- Replay the orders as they were picked. This is your baseline.
- Replay them on the same equipment with better batches, routes and slotting.
- Replay them with each equipment option, for example a goods-to-person zone for the fastest SKUs, or a different cart for some tours.
- Compare labor hours as well as meters, and price them at your own labor cost.
Step 3 shows what your current equipment can still do. Optioryx customers see 30 to 55% less walking when order clustering and pick path optimization run together. At a MedTech distribution center that picks with trolleys, grouping orders that share aisles on the same trolley came first. Route optimization and double-stacked trolleys followed, and picking productivity went up 27% with 54% less walking, as the MedTech customer story shows.
Steps 3 and 4 together decide the business case. Payback is the investment divided by the yearly saving. As an example with assumed numbers: a 400,000 euro system that saves 100,000 euros a year pays back in 4 years. If better batching on the current carts already removes 40,000 euros of that walking, the system adds only 60,000 euros a year, and payback moves to almost 7 years.
How much of the floor you need to model decides between a full digital twin and a lighter simulation. Bart explains which approach fits which warehouse.
This replay is how most projects with Optioryx Pulse start, on uploaded order history and a layout. Pulse plans batches, routes and slotting for the manual picking, on top of the WMS you already run, and can plan each tour for the cart or equipment type that fits it. Inside a goods-to-person system the system's own software keeps that job, so in a mixed site Pulse covers the zones where people walk.
Questions?
There are two families. Picker-to-parts systems keep people walking to the stock, with a cart and scanner, voice picking, pick-to-light or collaborative robots that carry the totes. Parts-to-picker systems bring the stock to a fixed station through goods-to-person storage, sometimes with robotic piece picking.
The three usually meant are picker-to-parts, where the picker walks to the location; parts-to-picker, where storage brings the item to a station; and fully automated picking, where machines also do the pick. Most warehouses use the first, and many combine it with an automated zone for their fastest items.
It depends on the order profile. Many small orders with overlapping items pick fastest in batches or clusters on one cart, on a route planned for that batch. Long orders suit single-order or zone picking. To know for your site, replay your own order history through each option and compare the hours.
No. Changing the method, the routes or the slotting can run on top of the WMS you already have, through an optimization layer that hands the planned work back to the WMS. New equipment such as goods-to-person storage does need an integration with the WMS, and that is usually part of the project.
Automation pays best with high, steady volume of small items that fit its bins and a shortage of people. Before you compare, take out the walking that better batches and routes would remove on your current setup. The business case should start from that improved baseline.