Your margin isn’t lost at sea. It’s lost in your office

It’s half past seven on a Thursday evening. The office is almost empty. One person is still there: the best person in operations, the one who knows every carrier’s surcharges by heart and spots a rate that doesn’t add up at first glance. And what is she doing? Checking invoices, one by one, copying numbers from a PDF into a spreadsheet. She’s not still there because she’s slow. She’s there because she’s the best, and that’s why the job lands on her desk.

We’ve seen this scene many times. We spent years on that side of the counter: at cargo airlines, at freight forwarders, managing international operations and in tech for the industry. And whenever the conversation turned to where the margin was leaking, it always went to the same places: freight rates, carriers, the price war.

Nobody looked at that person in the office at half past seven. Now Accenture’s latest report on supply chain and international trade operators has put numbers to what many of us suspected.

Profitability no longer rises and falls with the cycle. It stays down

The Accenture Research report Reinvention of Freight Forwarding (2026) starts from an uncomfortable fact. Between 2022 and 2025, industry revenue fell by an average of 6% a year, and the average EBIT margin stood at 8.5% in 2025, 10.5% lower than the year before.

One example explains it better than any chart. A large international operator lost around 30% of its EBIT on a revenue drop of just 5%. If you sell a little less and earn a lot less, the problem isn’t what you sell. It’s what it costs you to deliver it.

For years the industry’s answer has been to wait for the upturn. The report says, quite clearly, that this time it won’t come.

50-60% of hours burned on tasks that need no judgment

This is the figure that struck us most: between 50% and 60% of working hours in foreign trade and logistics operations go on tasks that AI can already automate today.

50-60% of a supply chain operator’s hours go on tasks AI can already automate.

This isn’t a vision of the future. Accenture breaks it down by area:

• Destination terminal operations: 59%
• Planning: 53%
• Pre-shipment documentation: 53%
• Customs: 52%
• Analytics: 51%
• Invoicing: 50%

Put another way, for every two hours your company pays for (whether you’re an importer, an exporter or a logistics service provider), one goes on keying in data, matching delivery notes, cross-checking rates by hand or wrestling with an invoice that contains an error.

And those errors are expensive. According to the same report, between 15% and 20% of the documents entering daily operations contain keying errors. On top of that, between 2021 and 2024, more than $10 billion was overbilled in freight charges and surcharges, and only a tiny share is ever recovered, because checking every line of every invoice from every carrier by hand is close to impossible.

The five processes where your time (and margin) goes

The report points to five processes that concentrate most of the manual work. You’ll probably recognize your own office in more than one of them:

1. Quoting and RFQs. Around 6% of lanes account for 80% of activity, and yet the same lane is quoted by hand, again and again.
2. Documentation and customs. This is where the 15-20% error rate shows up. Every error means a correction, a phone call or a delay.
3. Carrier selection. Decided by email, phone and the memory of whoever has been there longest.
4. Invoicing and rates. The quietest leak of all: what isn’t checked isn’t claimed.
5. Systems that don’t talk to each other. The TMS on one side, the ERP on another, each agent’s spreadsheet on a third, and someone acting as the bridge. The result: two to three days of delay in information.

That fifth point is the one we see most often, day to day. It isn’t that the tools fail. It’s that each tool is an island, and the person connecting them is, once again, the best person on your team. We cover this in more detail in our guide to ERP integration.

The risk isn’t failing to digitalize. It’s digitalizing halfway

There’s another figure worth keeping in mind. According to Accenture, only about 8% of international operators are true digital leaders, compared with 20% across logistics as a whole. And 76% admit that not investing in digital puts their business at risk.

In other words, almost everyone knows they have to do it, and very few have done it well. The most common outcome is halfway digitalization: one piece of software for each task, a project that dragged on forever and a tool nobody ends up using because “I’m faster with the spreadsheet.”

What we’ve seen ourselves: 90,000 invoices, 5,640 hours

Reports are useful, but we prefer real-world numbers. One of our clients handled 90,000 invoices a year. Their team spent 7,500 hours checking them: amounts, surcharges, rates. Skilled people working like robots.

We automated the process. Today 94% of those invoices are processed automatically, and the team spends 1,860 hours on the only thing that really needs a human eye: the exceptions. The result, in three figures:

• 5,640 hours freed up every year.
• €141,000 in annual savings.
• The investment paid for itself within a few months.

And nobody lost their job. Those hours went back where they belong: customers, lanes and growing the business. As a simple rule of thumb, every 10,000 invoices a year you automate frees up roughly a third of a full-time employee. We explain the process step by step in our guide to automated invoice processing.

And what about your P&L?

Combining Accenture’s data with our real case, this is the estimate by process:

Add it all up and an operator with the industry’s average margin (8.5%) could reach between 11.7% and 14.1%.

11.7-14.1% estimated EBIT margin for a supply chain operator that sits at the industry average today.

There’s another way to look at it: out of every 10 people who spend their day on these five processes, between 4 and 6 would have time freed up to sell, serve customers better and grow their accounts.

Where to start (without losing your mind)

You don’t need to automate everything at once. In fact, it’s better not to. This is what works for us:

1. Start with what hurts most. It’s almost always invoicing or documentation. That’s where the return shows up first and where the team feels the relief from week one.
2. Get organized before you automate. If the data is messy, automation only helps you make mistakes faster. Structure first, then the machine.
3. Make your systems talk to each other. Often you don’t need another piece of software; you need the one you already have to stop being an island.
4. Measure with two numbers, not one: how much margin you recover and how many hours you give back to your team. If you only look at one, you miss half the story.

Back to half past seven

The person from the beginning doesn’t need a machine to replace her. She needs the machine to take the repetitive work off her plate so she can do what she does best. For us, automation has always been about that: giving time back to people with judgment.

Imagine that same Thursday a few months from now. The invoices have been checked automatically throughout the day. The system has flagged three that don’t add up, and she has resolved them in twenty minutes, because that’s where her experience counts. At six she shuts down her computer. First thing tomorrow she has a meeting with a client she’s been meaning to call for weeks.

She hasn’t changed jobs. She’s got her real job back. And the company hasn’t hired anyone, but it has more capacity than ever.

And this is only the first step

In this article we’ve focused on what the Accenture report covers: automating the manual tasks that eat into margin. But don’t come away thinking that’s where it ends. That’s where it starts.

When an international trade company or a logistics operator asks us about AI, we always say the same thing: you work your way up to AI agents; you don’t jump straight to them. And you get there step by step:

1. Order. Clean data and clear processes. Without this, nothing that comes after works.
2. Automation. The repetitive work moves to the machine: invoices, documents, quotes.
3. Connection. Systems start talking to each other and information flows without anyone acting as a bridge.
4. AI agents. On that foundation you can have agents doing real work: preparing a quote, reviewing a customs file or spotting an invoice that doesn’t add up and alerting the right person. Always with someone with judgment supervising.

Companies that try to jump straight to step four usually end up with a beautiful demo that doesn’t work with their real data.

We’ve standardized this path, and not by chance: we’ve spent many years on both sides, in the industry’s operations and in technology. We know what hurts in an operations office and how to automate it without the implementation dragging on. Each step pays for itself before you take the next, so by the time you reach agents, you’re on solid ground. We look at what AI can already do in this area in AI in logistics.

How much margin are you losing?

That’s the question that really matters, and no report can answer it for you, not Accenture’s and not this article. Your numbers can.

So here’s a simple offer: we’ll run the numbers with you in 10 minutes. We only need three figures you know by heart: annual revenue, invoices issued per year and headcount in operations. You’ll walk away with two figures for your company: how many points of margin you could recover and how many hours you could give back to your team. No 40-slide deck, no commitment.

If it turns out it isn’t worth it, at least you’ll know where the leak is. And that’s worth a lot more than 10 minutes.

Get in touch and we’ll run the numbers together →

Warm regards,
Alexandra Escanero
Founder of Kaiona Tech

P.S. If you still think your margin is being lost at sea, try this: next Thursday at half past seven, take a walk around the office and see who’s still there and what they’re doing.

See you at Logistics & Automation Madrid. On November 11 and 12 we’ll be at IFEMA, stand 4B19 (hall 4), running this calculation live with every professional who stops by. Register for free here and get in touch to book a slot.

Source: Accenture Research, “Reinvention of Freight Forwarding” (2026). The margin impact estimates combine the report’s data with real Kaiona Tech cases and are indicative.

Key takeaways

• According to Accenture, between 50% and 60% of working hours in foreign trade operations go on tasks that AI can already automate.
• The industry’s average EBIT margin fell to 8.5% in 2025: the problem lies in the cost to serve, not just in what is sold.
• Invoicing and rates is the process with the biggest impact: automating it could add between 2.2 and 3.7 points of margin.
• You reach AI agents in stages: order, automation, connection and then agents.