Automation in a 3PL: How Do You Justify It When the Contract Is Three Years and the Client Might Leave?

Future Fulfillment in Melbourne is one of Australia's fastest growing ecommerce 3PLs, and it sorts more than 1,000 units an hour across 100 destinations on a Mushiny Minisorter. Its client mix changes constantly. That is the situation a 3PL automates into.

A retailer automates against its own ten-year volume forecast. A 3PL automates against a contract that may be re-tendered in three years, for a client whose SKU profile it did not choose and does not control. This article sets out what a 3PL can rely on instead, which parts of a Mushiny system are easy to change and which are not, and what the risk looks like if the client leaves. The examples come from the 3PLs on our reference sites page, and the main system is shelf-to-person.

Why is the normal automation case harder for a 3PL?

The normal case needs a volume forecast, a stock profile and a payback period. A 3PL controls none of them the way an owner does. Volume and range belong to the client, and the payback has to fit inside a contract someone else can decline to renew.

Mushiny's public position is that, after analysis and sizing, goods-to-person and sortation systems typically land at a return of under three years. That figure depends on how efficient the existing operation is, and it is the shape of the answer, not your number. Beside a three-year contract it leaves no margin for the client leaving early.

So the question changes. For a 3PL it is less "what is the fastest system?" and more "what happens to this investment when the client changes?"

What should a 3PL weigh differently from an owner-operator?

Does reconfigurability matter more than throughput?

For a 3PL, often yes. Peak rate tells you what the system does for the client you have today. Reconfigurability tells you what it does for the next one.

A shelf-to-person system is built from robots and shelf pods standing on an open floor. There is no conveyor and no fixed track. Mushiny's robots work out their own route at run time, and the pods can hold small and bulky items in the same system. The grid robots read QR markers on the floor to know where they are, and they work inside a fenced grid that people stay out of (our AMR article explains why that is faster and cheaper than SLAM). A marker tells a robot where it is, not where to go. The robots still plan their own routes and negotiate with each other.

A change in stock profile is a change in what is on the pods, not a change to a fixed machine. If the racking moves, the grid is laid out again and remapped. It is a job, but a defined one.

Compare that with a structure built around one machine. Our ASRS article says it plainly: a fixed structure built for one machine is hard to relocate, and if your lease or layout is uncertain a floor-based robot system is easier to change. A tall tote store such as MIX Pro earns its place where height and density are the constraint, but it is a gantry on rails inside racking up to 12 metres high. That is a commitment to the building, and a 3PL should price it as one.

Can the system move to another client or another site?

Yes, and we do it ourselves. Our Chatswood Tech Centre has been taken apart, trucked to trade shows around the country and reinstalled in place more than once. That is a small system and a customer site is a larger one, but nothing about the mechanism changes with scale. The robots and pods are separate units, the system is conveyor-free, and the grid is laid out and remapped wherever it lands. The same T-series robots run at sites from 5 robots to 205.

Where to be careful:

• It is a project, not an afternoon. The robots and pods move easily. Laying out and mapping a new grid takes planning, so budget it as a project rather than a removalist job.
• The QR markers are on the floor. They belong to the grid you lay out, so a move means laying out and mapping a new grid, not unplugging an old one.
• Gantry racking, such as MIX Pro, is a fixed structure. Treat it as part of the building.
• Sorters are sized to a site. A 3D Sorter MAX has its destination count set by machine length, so it is built for the building it goes into.

Re-pointing a system at another client inside the same building is a more realistic claim than moving it. The robots do not care whose stock is on a pod. They go where the software tells them.

Can you buy it in steps instead of all at once?

Yes. The systems are modular: you add robots, pods and storage as volume grows rather than re-platforming.

The reference sites show how wide the range is. Alloys runs 5 T6-800 robots against 100 pods. Coca-Cola in Sydney runs 16 against 114. Winc runs 107 against 1,421 pods and Munro Footwear Group runs 205 against 5,239. Those four are owner-operators, not 3PLs, but they show that the same system covers a very small start and a very large one. Starting small does not mean starting again later.

The sorters scale the same way. GPA Logistics in the USA, a third-party logistics provider handling consumer goods across B2B distribution and direct-to-consumer ecommerce, doubled its Mushiny H-shaped sorter from 80 to 160 sort destinations and runs about 2,000 units an hour.

Storage and speed get sized together, and stepping up is only cheap if the first step was laid out with room for the next. Ask for that in the design. For MIX Pro, throughput is raised by adding gantries and transfer robots, but the racking is the part you commit to on day one.

Can one system serve more than one client?

Our systems connect to your WMS, WCS, ERP and MES rather than replacing them, and the robots take their jobs from your warehouse management and order systems. Inventory ownership is a record in your WMS. A robot fetches whatever the software sends it for.

So the answer is yes to both. One grid can hold several clients' stock, and so can one pod. Ownership is a record in your WMS, not a physical boundary in the grid, and a robot has no idea whose carton it is carrying. For a 3PL that is the point: you are not buying one system per client, you are buying capacity and allocating it.

What still needs deciding is whether you *want* stock mixed. Bring these questions to sizing.

1. Does any client need its stock physically separated, rather than separated in the WMS?
2. Does your WMS already track stock by owner at location level?
3. Does any contract or compliance rule require segregation in writing?
4. On a sorter, do orders from different clients run in the same batch or in separate ones?

Future Fulfillment and GPA both serve a mix of clients, and Future Fulfillment's case study calls a changing client mix the case a manual putwall handles worst. Sortation suits a 3PL for the same reason the grid does: the machine is indifferent to whose order it is sorting.

Can automation be something a 3PL sells?

A 3PL that can pick faster, sort accurately and pack without queues can offer a client things a manual site cannot. The inputs are shelf-to-person picking at up to four times the rate of manual pick-and-walk lines per hour, 99.9 per cent sort accuracy, and a Minisorter that handles returns as well as outbound.

We have no cut-off figures from any reference site. The argument is simple. How late you can promise a same-day cut-off depends on how fast you can pick, sort and pack. If automation moves it later than a manual site can offer, it wins work that site cannot bid for.

Packing counts too. The PolyPacker bags and labels at up to 1,500 bags an hour. Rental options are available in Australia for it through our partnership with Omni Group. Rental is not offered across the other systems, so ask us if that matters to your case.

What happens if the client leaves?

If the system was shaped around that client, you own a system shaped around nobody. That is the risk, and it has three forms.

Stock profile. If the pods, tote sizes or sorter bins were configured around one client's product, the next client's product may not fit. Sorter item envelopes are fixed by machine class, so check the next client's range against them.

Volume. A fleet sized for the departed client's peak is oversized for the next one. Do not build the business case on stripping robots out and redeploying them.

Layout. A fixed structure such as a gantry store stays where it is. A floor-based grid can be laid out again, at a cost.

You can reduce these risks without removing them. Keep the first step small. Design the layout for the second client, not only the first. Favour configurations with the widest SKU envelope, such as pods that carry small and bulky stock together. Do not put the largest fixed commitment behind a single client.

What should you bring to the first conversation?

Bring your contract term, each client's tenure and SKU range, and your peak order lines per hour. Add the question that matters most: what does it cost to change this system for the client after next?

Take them to the Chatswood Tech Centre, the only place in Australia with Mushiny's full range running live. You can run a client's product through it, and visits are free.

Frequently asked questions