How to calculate the ROI of factory automation

Start with the losses, not the technology

The return on factory automation is not a property of the technology — it is a property of the loss it removes. Before modelling any investment, write down where the operation actually loses money: unplanned downtime, scrap and rework, overtime, and skilled hours lost to manual admin. Automation only earns a return where it removes one of those losses, measurably.

Calculate each benefit independently

The single biggest mistake in automation business cases is double-counting. The same recovered hour gets counted as extra output and as saved labour; overtime savings get counted twice. Calculate each stream on its own and keep them from overlapping:

The realisation factor matters. Not every recovered hour becomes cash. A recovered hour is only cash if it removes overtime, removes temporary labour, is redeployed to productive work, or justifies a genuine headcount change. Counting every saved hour as full payroll saving is the fastest way to build a business case that never materialises.

Value downtime at your contribution margin per productive hour, not at revenue. If you don't know your margin, treat downtime as avoided idle-labour cost instead — a smaller, honest number — rather than inventing output you may not be able to sell.

Only count extra output if you can sell it

Extra capacity is not profit. Additional output only counts when there is real demand, the materials and downstream capacity to make and finish it, and a known contribution margin. If any of those is missing, the honest figure for extra output is zero.

Turn annual benefit into a decision

Once you have a defensible net annual benefit (gross benefit minus recurring software, support and maintenance costs), the investment maths is simple:

• Simple payback = investment ÷ net annual benefit
• Three-year net impact = net annual benefit × 3 − investment
• Three-year ROI = three-year net impact ÷ investment

Then stress-test it. If the benefit lands 20% below your estimate, does the payback still make sense? If a project only works on the optimistic assumption, it is a risk, not a plan. And if the net benefit is negative, say so — a transparent "this doesn't pay back" is worth more than an attractive invented number.

You can run exactly this logic, with every assumption editable, in the GlobXAI ROI calculator.