What Warehouse Automation Costs in Australia, and How to Work Out the Payback

From 1 July 2026 the lowest rate a warehouse employee can be paid under the Storage Services and Wholesale Award is $27.08 an hour. That is the Storeworker grade 1 rate, before superannuation, leave and every other on-cost. It is the first number in any Australian automation business case, because most of what a system saves is labour.

This article does not quote a price for any machine. Cost depends on the system, the volume, the building and the integration, and a figure printed in a general article would be wrong for nearly every reader. What it gives you is the method: which lines go into the case, the Australian inputs for the two biggest ones, and what moves the answer. If you want the cost side from Mushiny, a sizing study on your own order data produces it, for a system such as shelf-to-person.

What actually goes into an automation business case?

A payback calculation compares two cash flows. One is what the operation costs to run today. The other is what it costs to run with the system, plus what the system costs to buy, install and keep running.

Most cases are built from five lines:

1. Labour saved. Headcount you no longer need for the same output, or output you can reach without hiring.
2. Floor space avoided. Square metres you can give back, sublet, or not lease when you grow.
3. Accuracy and returns. Fewer wrong orders, fewer credits, fewer reshipments.
4. Capacity gained. Volume you can now handle at peak without temporary labour or a second shift.
5. System cost, all in. The machines, plus the lines covered further down that people leave out.

Lines 1 and 2 can be built from Australian public data, and the next two sections do that. Lines 3 and 4 depend on your own order history. Line 5 comes from a quote.

How do you build the labour line from the award?

Start with the Storage Services and Wholesale Award, MA000084. From the first full pay period on or after 1 July 2026, a Storeworker is paid from $27.08 an hour at grade 1 to $29.61 an hour at grade 4. On a 38 hour week that is $1,029.10 to $1,125.30 a week. Across 52 weeks it is roughly $53,500 to $58,500 a year per full-time position.

That is a floor, not a cost. Treat it as the base of a build-up:

• Add superannuation.
• Add leave, leave loading and public holidays.
• Add workers compensation, payroll tax, penalty rates and overtime where they apply.
• If you use agency staff, replace all of the above with the agency's charge rate, which includes their margin.

The number you want is the fully loaded cost of one hour of productive picking, packing or sorting at your site. Your payroll team can give it to you, and it will be higher than the award rate.

Be exact about what leaves. Automation rarely removes a whole function. Someone still inducts items, replenishes stock, clears completed orders and handles exceptions. The saving is the difference between the hours you spend today and the hours you spend afterwards, valued at your loaded rate. Count both sides. A case that subtracts nothing for the people who stay will not survive a finance review.

Jingyuan Supply Chain in Xi'an cut its picking team from 20 people to 7 and returned the cost of its system inside a year. Mushiny describes that as faster than typical, and says it reflects how manual the site was beforehand. It shows what a very manual starting point can do. It does not tell you what yours will do.

Why does face rent overstate the floor saving?

Floor space is the second biggest line, and it is the one most often inflated. The usual method multiplies the square metres you free up by the rent per square metre in your lease. That uses face rent. In a market where landlords compete for tenants they pay incentives such as rent-free periods and fit-out contributions, so the rent you effectively pay over the term is lower than the rent on the page.

Prime industrial leasing incentives in the first quarter of 2026, as reported by Knight Frank, were:

For every $100 of face rent, a tenant in Western Sydney effectively pays about $78 once the incentive is spread across the term. In Melbourne it is about $79.50, in Brisbane about $87.60, in Perth about $90.40 and in Adelaide about $91.10. Those figures are simple arithmetic on the percentages above, and your own lease will differ.

So value avoided floor space at net effective rent, not face rent. In the two largest markets the face figure is up to about a fifth too high, and a finance team that knows industrial leasing will check.

Two cautions on this line:

• It only counts if the space turns into money. Freed floor you cannot sublet, or space you would never have leased, is worth nothing to the case. The strongest version is the lease or expansion you avoid.
• If you own the building, the relevant figure is the value of the best alternative use of that space, not a market rent you do not pay.

What costs do people forget?

The machines are the part everyone remembers. These lines are the ones that make a case look better than the project turns out.

• Integration. The system has to talk to your WMS, WCS or ERP. Mushiny systems connect to these rather than replacing them, but the connection is still work, and some of it falls on your side.
• Training and ramp-up. Staff are not at full rate on day one. Include the weeks where output is below target.
• Support and maintenance. Service, spares and software, for every year of the horizon you are testing.
• Peak casuals you will still need. If your Christmas or promotion peak is larger than the system's design rate, you will keep hiring for the difference. Do not count the whole peak labour bill as saved.
• Disruption. Most Mushiny systems are brownfield installs into a working building, and several Australian sites have gone live over a single weekend. Even so, plan for some loss during the transition.

A payback worked out against a partial cost gets corrected later, in front of the people you showed it to.

What makes payback faster or slower?

Mushiny's published position is that payback depends heavily on how efficient your operation already is. After analysis and sizing, goods-to-person and sortation systems typically land under a three year return, and the PolyPacker bagging machine tends to pay back in under a year if the volume is there. Treat both as the shape of the answer, not your number.

Payback is faster when:

• Pickers spend much of their time walking. Mushiny's shelf-to-person page puts non-productive walking at about half of a picker's time on a typical manual site.
• The work is manual and the process has not been tuned, which is what Jingyuan had.
• The system runs more shifts. A fixed cost set against two or three shifts of avoided labour pays back sooner than one set against a single shift.
• Volume is high enough that the machine is rarely idle.
• You lease in a market where you can genuinely give back or avoid floor space.

Payback is slower when:

• The operation is already efficient, with a dense, compact pick face and little travel to remove.
• Volume is low or seasonal, so the machine sits idle for part of the year.
• Your constraint is somewhere else, such as the packing bench. Automating picking when packing is the bottleneck moves the queue and saves nothing.
• Your product falls outside what the machine handles, so a large share of volume stays manual.

Which input matters most?

Your own cost per order line. It turns every other number into dollars, and only you have it.

Cost per line is the total cost of picking, packing and despatching, divided by the lines you ship. Include labour at the fully loaded rate, the share of floor space the process uses at net effective rent, supervision, consumables and errors. Work it out for today, then work it out again for the automated process. The difference, multiplied by your annual lines, is the saving.

Everything above feeds that calculation. None of it replaces it. Two sites with the same headcount can have very different costs per line, because of how far their pickers walk, how many lines each order carries and how much of the volume is peak. Public benchmarks cannot see that. Your WMS can.

How do you take this to a CFO?

Show the calculation, not just the conclusion. Give the baseline from your own systems, labour at the loaded rate, floor space at net effective rent, and the forgotten costs above, each with its source. Then test it at normal volume, at lower volume, and with the saving cut by half. A case that still holds is stronger than one that only works on the best assumptions. Mushiny's reference page lists named installations you can read against your own operation.

To have Mushiny size it against your data, enquire here. The Chatswood Tech Centre is the only place in Australia with Mushiny's full range running live, and it is free to visit.

Frequently asked questions