In 2014, every startup wanted to be pure software. High margins, high multiples, endless scale.
Flexport did the opposite.
They didn't build a layer of software on top of the logistics industry. They went into the industry itself and started running the operations. That made them a freight forwarder with very good software underneath, in a way.
I asked Sanne if they ever considered going plain vanilla SaaS. His answer: "No. That was a clear choice in 2014."
The thesis: you can't automate what you don't run
Sanne's reasoning was simple. Logistics has enormous amounts of data, but almost all of it is unstructured. Because of that, processes can't be automated, and customers get a poor experience.
Putting a software layer on top doesn't fix that. So Flexport decided to deliver the service itself, and use that position to structure the data from the inside.
"Before you can automate, you have to digitise," Sanne said. For about ten years, they broke global trade down into small atomic tasks and digitised each step, so a machine could understand it.
The long-term goal was always the same: an operating system for global trade.
Not a SaaS company, a full stack
When I asked how to define Flexport, Sanne described a stack of four layers:
Infrastructure: people around the world who solve problems when the software can't.
Data: millions of transactions, plus Atlas, their world model of logistics assets and schedules.
Operating workflows: the processes that move freight.
Software on top: the interfaces for customers, ports and truckers.
The point of doing the whole stack is to lower costs, improve the customer experience, and connect everyone to the network.
The price of being different
That choice wasn't free. Investors looking at Flexport saw infrastructure, services and a lot of people, and said: this isn't SaaS, we don't understand it. Fundraising was harder. Some employees even left for SaaS companies because it seemed easier.
Why the "handicap" became the moat
Now AI changes the maths. Ten years of structured data and digitised tasks are exactly what agents need.
Flexport grew 28% year to date without increasing headcount. Sanne said that classic automation could handle about 64% of tasks. With agents, they believe 95% is possible, and they're at 81% today.
Meanwhile, Sanne sees competitors stuck on one side. Traditional forwarders have the execution, but not the data. Transportation software vendors have software, but no execution layer. Flexport has both.
What this means for founders
Owning the operation can be a moat, not a weakness.
Structure the messy part of your market. That's where the value accumulates.
Margins today aren't the whole story. The hard, unglamorous layer is often the one nobody can copy.
Maybe the real question isn't whether you're a SaaS company. It's which part of the stack you own.
Listen to the full conversation with Sanne Manders on the We Love SaaS podcast.
