AMRs: how to calculate ROI, TCO and payback in intralogistics

The business case for an AMR nearly always breaks at the same point: the line where someone wrote "one operator fewer." The operator usually stays. That person spent forty percent of the shift moving material, and the other sixty percent doesn't leave with the robot. Annual savings drop to less than half of what was budgeted, payback becomes almost three times longer, and the discovery comes when the vehicles are already on the floor. A calculation that holds up starts from the hours that are truly repetitive, books the costs that come back every year (energy, maintenance, batteries, software licenses) and compares the AMR with the realistic alternative, which is almost never standing still.

Forklift, AGV, AMR: three different cost profiles

A manned forklift is cheap in capital and expensive to run. The dominant cost is the driver's hourly rate multiplied across every shift, and in return the company gets total flexibility on non-standard tasks. A path-guided AGV carries a middling capital cost plus the infrastructure laid in the floor. It performs very well on very high, unchanging flows and becomes slow and expensive whenever the layout changes. An autonomous mobile robot with SLAM navigation costs more per vehicle but needs no fixed infrastructure, reconfigures in software and adapts to routes that change. The technical differences between the two automated vehicles are covered in our AGV vs AMR comparison.

That leads to a practical rule. An AMR rarely beats the forklift on occasional moves, while it does beat it on repetitive, predictable flows spread over several shifts, because that is where handling labor, multiplied by the shifts, becomes the cost that automation goes after.

The savings that hold up in front of a finance director

The big saving, the one that carries the calculation on its own, is a single item: the hours people spend moving material instead of working on it. It has to be measured on the share of the work that can really be automated, not on the whole shift. Someone who moves material forty percent of the time frees up four tenths of a person, and everything else is built on that figure. The machine adds one item of its own: an AMR also works the second and third shift with no breaks or absences, so the same fleet runs more cycles per day and replaces more hours.

Below that line sit benefits that are just as real but harder to book, and they should be declared as estimates. Damage from manual maneuvering (struck goods, uprights to straighten, the aisle closed while it gets fixed) lands on no cost center in many warehouses, yet it comes back every month. A repeatable path at controlled speed reduces it, and six months of counting is enough to know by how much. Errors follow the same logic, with a tracked mission, delivery to the right point and a move recorded in the system. Then there is the item most companies discover late, the backs of the people doing the lifting. EU-OSHA ranks musculoskeletal disorders among the most common work-related health problems in Europe, with a cost in lost days and indirect expenses far larger than the compensation itself. The injury side is covered in our article on AMRs and workplace safety.

Simple payback is the total investment divided by the net annual saving. It works as a first screen and is not enough to decide, because it ignores recurring costs in later years and the time value of money. A solid decision moves on to multi-year total cost of ownership and, when the amount justifies it, to a discounted cash flow with NPV and IRR.

What goes into TCO beyond the vehicle price

Total cost of ownership over the full service life is the yardstick for comparing an AMR, an AGV and the current setup, and the purchase price is only part of it.

Two items are underestimated almost every time. The first is battery life: batteries have a limited number of cycles and get replaced during the vehicle's life. The charging strategy decides both how long they last and how available the vehicle is, as explained in our piece on sizing AMR fleet charging. The second is infrastructure, where the AMR has a structural edge over a path-guided AGV because it needs no wires, magnets or tape and forces no downtime when the layout changes.

A worked example with the assumptions on the table

There is no payback range that holds for everyone, and we don't publish one without a source to back it. What we can show is how the number moves. To keep the example free of prices that vary from case to case, every item is expressed as a multiple of the full annual cost of one handling operator, which we call AC. The assumptions are ours, written to show the method, and should be replaced with your own data.

The correct calculation starts from the hours: four people at forty percent make 1.6 full-time equivalents, and at seventy-five percent that becomes 1.2 AC of avoided cost every year. Take away recurring costs and 1.0 AC net remains, so simple payback is three years. The same flow on three shifts, with six people involved and recurring costs up to 0.25 AC, frees 2.4 full-time equivalents, that is 1.8 AC of avoided cost and 1.55 AC net: payback drops to about two years on the same investment. The wrong calculation, the one that counts four whole people as savings, would show 2.8 AC net and a payback of just over a year. That is the number acceptance testing will never confirm.

The example also shows where to act before signing. Shifts are the strongest lever, the automatable share has to be counted on missions rather than on people, and the conversion factor depends on organizational choices (hires avoided, temporary labor, overtime, redeployment) that belong to the company, not the supplier.

Your payback is decided by the week you counted the missions

Then there is a variable that appears in no quotation. Mission data (how many moves, between which points, at which hour) is almost always collected over a single week, because a week is what the warehouse will give you. An ordinary week, though, does not contain the peak. A fleet sized on that week passes its acceptance test, meets the specification and then stalls in November, because the test measured exactly what it was asked to measure. The method for sizing a fleet on the worst hour, with a worked calculation, is in our article on evaluating an autonomous forklift and AMR project. For automated warehouses the request is the same, twelve months of measured movements, as explained in our guide to choosing an automated warehouse.

Integration time with WMS and MES doesn't take benefits away. It delays them, and in a payback calculation a delay weighs as much as a missing saving. To be defensible, the comparison has to be made against the real alternative: an AGV, two extra operators, an evening shift. It is the comparison the finance director will make anyway, so it pays to make it first.

The technical choices that move the result

Design choices have a direct effect on the numbers. The vehicle has to be sized on the real load (pallets, roll cages, bins, totes), because an oversized vehicle is idle capital and an undersized one leaves missions it cannot run. SLAM navigation removes fixed infrastructure and allows reconfiguration in software. Magnetic guidance remains an option for very rigid flows and penalizes TCO whenever the layout changes. Good fleet management software optimizes traffic, charging and priorities, which raises vehicle utilization and the return with it.

Interoperability is the item you pay for three years later, when the project is already considered a success. The VDA 5050 specification, which reached version 3.0.0 in March 2026, standardizes how orders and states are exchanged between fleet control and vehicles from different makers. Requiring it in the purchase order costs one line in the specification. Not requiring it costs the second tender, the one where the only supplier able to bid is the incumbent. On compliance, the vehicles must meet ISO 3691-4:2023, in its harmonized form EN ISO 3691-4:2023 where CE marking is concerned, and from January 20, 2027 Regulation (EU) 2023/1230 replaces the Machinery Directive 2006/42/EC. A non-conformity found at acceptance is a hidden cost that can stop the project.

What to send us for an assessment

To build a business case with written assumptions we need the following data, most of which the company already has:

• flows to automate, with origins, destinations and distances;
• missions per shift and per hour, with the worst hour kept apart from the average;
• number of shifts and working days per year;
• people involved and the share of their time spent on transport;
• load type, weights and dimensions;
• handling-related damage, errors and stoppages over the last six to twelve months, if tracked;
• systems to integrate (WMS, ERP, MES) and layout constraints.

You can send them by email to info@pitech-solution.com or on WhatsApp. The available technologies are described on the pages about robotic forklifts and AMRs for pallet handling and industrial automation and material handling robotics.