Goods-to-Person Systems

Walk any conventional distribution centre during a busy shift and you will see a great deal of motion and surprisingly little value being added. Pickers push carts up and down aisles, scan a location, take one or two items, and move on to the next location that could be a hundred metres away. They are working hard. They are also spending most of their paid time walking, and walking does not put a single item into a customer order. Goods-to-person systems attack this waste directly by inverting the whole model: the picker stays at a station and automation brings the storage to them. This guide sits under the broader warehouse automation complete guide, and it drills into the one idea that changes picking economics more than any other.

The message up front: goods-to-person does not make picking faster by making pickers work faster. It makes picking faster by deleting the travel between picks. That single structural change is where almost all of the throughput and accuracy gains come from, and it is also why the technology only pays on operations with the pick volume and the density to justify the capital.

1. The walking problem in picking

In a traditional person-to-goods operation the worker travels to the inventory. Study after study, and every time-and-motion exercise I have run on a warehouse floor, lands on the same uncomfortable figure: travel typically accounts for somewhere between forty and seventy percent of a manual picker's working time. The item pick itself, the reach-grab-place that actually fulfils the order, is a small fraction of the shift. Everything else is the picker moving their body and their cart from one location to the next.

This matters because travel time scales badly. As you add more products, the storage footprint grows, and the average distance between picks grows with it. As order volume climbs, you add more pickers, and now they congest the same aisles. The operation gets slower per unit precisely when you most need it to get faster. Manual picking has a ceiling, and that ceiling is set by geometry and human walking speed, not by how motivated your team is.

There are conventional ways to shave travel: slotting fast movers near despatch, batch picking multiple orders in one trip, zone picking so each worker covers a small area, pick-to-light to cut the search time at the shelf. All of these help, and none of them remove the fundamental problem. The picker is still walking. Goods-to-person is the only model that eliminates the travel entirely rather than optimising around it, which is why it represents a different category of improvement rather than an incremental tune.

2. What goods-to-person means

Goods-to-person, often written GTP, describes any picking model where automated storage and transport bring inventory to a fixed picking station, and the picker never leaves that station. The order picker stands at an ergonomic workstation. A shuttle, a crane, a robot or a conveyor delivers a bin, tote or tray containing the required product to the station. A light or a screen tells the picker exactly what to take and how many. The picker confirms the pick, the container is whisked away, and the next one is already arriving. The picker becomes a stationary, high-throughput node instead of a mobile searcher.

The diagram below contrasts the two models. On the left, a person walks the aisles to reach the goods. On the right, automation carries the goods to a person who stays put.

The consequence of that inversion is not subtle. When the picker never walks, pick rates commonly rise from the fifty to one hundred lines per hour typical of manual picking to several hundred lines per hour at a well-fed goods-to-person station. The picker is doing the same physical act of reaching and grabbing, just without the dead time in between. And because the system presents exactly one item and one quantity at each cycle, the opportunity for a wrong-item error collapses.

3. The technologies that deliver it

Goods-to-person is a principle, not a product, and several distinct families of technology implement it. Each has a different sweet spot in terms of density, throughput, SKU size and cost, and choosing between them is where most of the engineering judgement lives.

Shuttle systems and cube storage. A dense grid or racking structure is served by fleets of small shuttles that run along levels or climb the structure to retrieve totes and deliver them to a lift, which passes them to the picking station. Cube-based automated storage, where robots move across the top of a stacked grid of bins and dig totes out on demand, is the best known modern variant. These systems achieve extreme storage density and very high throughput, which is why they dominate high-volume e-commerce and parts distribution. They are also the most capital-intensive option and the most disruptive to install.

Automated storage and retrieval systems (AS/RS). The older and broader family: crane-based aisles that store and retrieve loads automatically, in unit-load form for pallets or mini-load form for totes and trays. Mini-load AS/RS feeding a picking station is a classic goods-to-person configuration and has been in service for decades. It suits deep, stable inventory and heavier or larger cases that shuttle grids handle less gracefully. For the full picture on this family, see automated storage and retrieval systems.

AMR-based goods-to-person. Autonomous mobile robots drive underneath movable shelving pods and carry the entire pod to a picking station, then return it when the picks are done. This is the model that made goods-to-person famous in large e-commerce fulfilment. It trades some storage density for enormous flexibility: you can add robots and pods incrementally, reconfigure the floor without ripping out steel, and scale up for peak. AMR-based systems are the most adaptable and the least locked-in of the three. See autonomous mobile robots (AMRs) for how these platforms work, and the broader warehouse robotics explained for where they sit in the landscape.

The practitioner's read: shuttle and cube systems win on density and raw throughput per square metre, AS/RS wins on deep stable stock and larger loads, and AMR-based systems win on flexibility, phased investment and the ability to reshape the operation as demand changes. There is no single best answer, only the best fit for a specific profile of volume, SKU characteristics and how certain you are about future growth.

4. Goods-to-person versus person-to-goods

Laying the two models side by side makes the trade clear. Goods-to-person wins decisively on the operational metrics that scale with volume, and person-to-goods wins on the metrics that matter when volume is low or unpredictable.

The pattern in that table is consistent. Goods-to-person is superior on every operational quality metric and inferior on flexibility and upfront cost. That is the whole trade in one sentence: you convert a large capital outlay and a loss of surge flexibility into permanently lower labour per order and permanently higher accuracy. Whether that conversion is worth it depends entirely on how much volume you push through it.

5. Where it pays: high pick volume, dense SKUs

Goods-to-person is a fixed-cost machine that earns its keep by amortising that fixed cost across a very large number of picks. The economics are therefore ruthless about volume. A few conditions have to line up for the investment to return.

• High and sustained pick volume. The station and the automation are paid for whether they process a thousand lines a day or fifty thousand. Only high, steady throughput spreads that cost thin enough to beat the labour it replaces. Seasonal spikes on an otherwise quiet baseline do not justify permanent automation.
• Many small SKUs. Goods-to-person shines when you have thousands of distinct small items that would otherwise force pickers to walk enormous distances. Totes and bins suit small parts, consumer goods, pharmaceuticals, spares and e-commerce inventory. A handful of bulky pallet SKUs is the wrong profile.
• Each-level or piece picking. The model is at its best picking individual items or eaches into orders, which is exactly the profile of e-commerce and service-parts fulfilment. Full-pallet movement is better served by conventional handling.
• Constrained or expensive floor space. Because the storage grid needs no human walk-aisles, goods-to-person packs far more inventory into a given footprint. Where land or building cost is high, that density becomes a second, independent source of return.

The clean summary I give clients: goods-to-person pays where you have a lot of small things, a lot of orders, and not a lot of space. Hit all three and the return is compelling. Miss any one badly and the arithmetic starts to wobble.

6. The honest limits: capital and peak flexibility

Every serious evaluation of goods-to-person has to reckon with two limits that vendor case studies tend to soften.

Capital and commitment. These systems are expensive to buy, expensive to install, and in the case of fixed shuttle and crane infrastructure, expensive and disruptive to change once in. You are committing to a physical design and a demand assumption for years. If your business pivots, your SKU profile shifts dramatically, or your volume forecast was optimistic, you are left with a large asset sized for a reality that did not arrive. AMR-based systems soften this by being modular and relocatable, which is a genuine advantage when the future is uncertain, but they still carry real capital and integration cost.

The honest limitation: goods-to-person has a hard throughput ceiling set by the number of stations and machines you installed. A manual operation absorbs a surprise peak by drafting in temporary labour for a week. An automated operation cannot conjure extra shuttles overnight. If your demand is spiky and unpredictable, a fully automated goods-to-person floor can become a bottleneck at exactly the moment you most need capacity, and you end up bolting manual picking back on to cover the peak.

Peak flexibility. This deserves emphasis because it catches people out. The elegance of goods-to-person, its stable and predictable rate, is the same property that makes it inflexible. It runs at its designed rate beautifully and it does not run above it. Many of the strongest operations therefore run a hybrid: automate the dense, high-frequency core of the catalogue where the volume is reliable, and keep a manual person-to-goods zone for the long tail, oversized items, and surge capacity. Treating goods-to-person as an all-or-nothing choice is usually a mistake; treating it as the automated core of a blended operation is usually right.

7. Goods-to-person, the WMS and pick optimization

A goods-to-person system is only as good as the software directing it, and this is the part that gets underestimated in almost every business case I review. The physical hardware moves bins, but the decisions about which bin, to which station, in what sequence, batched with which other orders, are made by software. Get that layer wrong and an expensive machine delivers mediocre throughput.

The warehouse management system is the brain of the operation. It holds the inventory record, receives the orders, and decides how work is released. Sitting under or beside it, a warehouse execution or control layer choreographs the automation in real time. The WMS has to do several things well for goods-to-person to hit its numbers: batch orders so that each bin presented to a station satisfies as many order lines as possible, sequence deliveries so the picker is never waiting for the next tote, balance work across multiple stations so none sits idle, and keep the fast-moving inventory positioned where the automation can retrieve it quickest. For the fundamentals of this layer, see what is a WMS.

This is also where the integration challenge lives, and it is the challenge closest to my own work. A goods-to-person deployment is rarely a single vendor's monolith. It is a WMS, an execution layer, the automation controller, and usually the ERP that owns the master data and the order source, all of which have to exchange messages cleanly and in real time. The bin does not move until the message says move it, and the picker does not see the right instruction unless the data lined up across four systems. When these integrations are sloppy, the symptom is not a crash, it is a slow, mysterious underperformance where the machine is capable of far more than the operation ever extracts from it. The hardware is almost never the thing that limits a goods-to-person operation. The software orchestration and the system integration are.

The link back to the broader picture is worth making explicitly: goods-to-person is one equipment choice within a full automation strategy, and it only delivers when it is designed alongside the WMS, the slotting logic and the wider material flow. The warehouse automation complete guide places it in that context and covers the adjacent decisions that determine whether the whole system, not just the picking station, actually performs.

8. References

The figures in this guide, particularly the share of picker time spent travelling and the throughput ranges, are drawn from widely reported industry benchmarks and from time-and-motion observations in real distribution operations. Treat them as representative orders of magnitude rather than precise universal constants; every facility differs by layout, SKU profile and order pattern.

• Bartholdi, J. and Hackman, S., Warehouse and Distribution Science (open-access text), on picker travel as the dominant component of order-picking labour.
• de Koster, R., Le-Duc, T. and Roodbergen, K., "Design and control of warehouse order picking: a literature review," European Journal of Operational Research, on person-to-goods travel and picking system design.
• MHI and industry association material on automated storage and retrieval systems, shuttle and cube storage, and mobile-robot goods-to-person fulfilment.
• Vendor technical documentation on mini-load AS/RS, shuttle grids and AMR-based pod-to-station systems, cross-checked against the operational ranges cited above.

Final thoughts

Goods-to-person is one of the clearest ideas in warehouse automation because it fixes one clearly identified waste: the walking. Remove the travel and you remove the largest slice of picking labour, lift throughput several times over, and cut errors sharply, all without asking anyone to work harder. That is a rare kind of improvement, and on the right operation it is transformative.

It is also not free and not universal. It demands high steady volume, a dense small-item catalogue and the capital to commit, and it trades away the surge flexibility that manual picking gives you for nothing. The best operations do not choose goods-to-person against manual picking; they automate the reliable, high-frequency core and keep manual capacity for the tail and the peaks. Do the volume arithmetic honestly, invest as much attention in the WMS and system integration as in the hardware, and goods-to-person delivers exactly what it promises. Skip the arithmetic and you buy an elegant, expensive machine that runs beautifully at a rate your business never needed.

Related reading: Warehouse automation: the complete guide, Automated storage and retrieval systems, Autonomous mobile robots (AMRs), What is a WMS, Warehouse robotics explained.