TL;DR

• Torfs is Belgium's largest (and most beloved) shoe retailer, and just made its first acquisition abroad.

• Behind the stores is finance work that's high-volume, messy, and can't be wrong.

• They tried RPA and copilots like Claude, but none could keep up. Now, Pointer runs that work from start to finish (even exceptions).

• Their finance team used to spend its days copying data between systems and checking if items matched. Now, that time goes into growth.

Belgium's favorite shoe company

When I landed in Belgium for the first time, the border guard asked me why I was visiting. "Torfs," I told him.

"Torfs?" he said. "My entire family buys shoes there."

Torfs is the Trader Joe's of Belgian shoes: everyone has a happy memory of the place.

"We're really a loved brand," says Stijn De Knop, their CFO. "It's really about coming into the shop and feeling welcome."

That welcome has lasted over 75 years, with Torfs named Belgium's best employer fifteen (!!) times. And they aren't slowing down, with a recent acquisition and international expansion.

But every CFO knows that fast growth means more back office work. At Torfs, that all landed on a few desks.

During my first visit, Torfs' back office ran almost entirely by hand. Clerks filed paper orders into the RMS manually, copied data between systems on split screens, and checked thousands of supplier order lines a day against what they'd ordered, line by line, all day, every day.

Less than three months later, I stood in a transformed Torfs headquarters: software reconciled daily revenue from over 150 registers; programs checked tens of thousands of supplier confirmations, line by line; photos of shoe pairs automatically generated new SKUs - all work their team used to do by hand.

The 'before' picture resonates with almost every customer we talk to, and the pain has been obvious for years. So why hadn't anyone solved it for Torfs?

A decade of failed automation

Torfs recognized this problem early. Five years ago, they hired a team to build an RPA bot - software that clicks through screens in a fixed, programmed path.

It never stood a chance. Torfs' work has 15 exceptions for every rule: some depend on the day, others even the weather. So when automations took months to set up and couldn't handle the simplest cases, Torfs scrapped them. In Stijn's words, it was "more work than we could gain from it."

More recently, Torfs tried copilot tools like Claude and ChatGPT, too. They found the tools helpful for small tasks, like writing emails and editing spreadsheets.

But they're not built to book a new receipt every 10 seconds from over 150 sources across the country, and get every single one right, day after day.

Torfs tried robots and copilots, but the work still landed at the desk of Jill, their Chief Accountant, every morning. Just keeping up with demand meant spending resources on existing business, making expansion even more daunting.

Something had to change.

Stijn's bar: A to Z, or don't bother

We see this problem with every customer we work with: automation software isn't powerful enough to give the finance team its day back, because it can't complete the work from start to finish.

The dirty secret with existing automation software is that it handles the first 80%, and a human has to run the last mile. We've seen this time and time again, from the days of RPA and OCR to modern AI tools.

The invoice scanner reads the PDF, then someone has to fix what it got wrong. The AP software is ready to pay, but when numbers don't match, someone digs up the PO to check the terms. The reconciliation tool flags a gap, then someone spends all afternoon tracing it back to one store.

In other words, that last 20% is the part that takes all day. Stijn agreed completely:

"Processes have to be really automated from A to Z."

That's a core belief of ours, too. We never set out to build another automation tool, because automated work still needs someone to babysit it. We set out to eliminate the work entirely.

Torfs is moving off its custom systems to a new ERP, and it didn't wait for that migration to start winning. Here's how Torfs did it.

Getting to A to Z

1. Torfs' team teaches the job

75 years of selling shoes leaves dozens of rules trapped in people's heads, but those rules are second nature to the Torfs team. Where a traditional automation sounds the alarm over a register discrepancy, Jill knows that's just store staff borrowing cash for lunch.

So, Jill and her team needed to teach Pointer the same way they'd onboard a new hire. Otherwise, Pointer would fail at the first diversion from the status quo (defeating the entire purpose of automation).

It didn't take long for Pointer to understand all of those rules. In Jill's words: "I had to see that they really understood the process, and that came really quickly."

2. Jill puts it to the test

But knowing the process on paper and actually doing it are two very different things. Jill and her team weren't going to hand Pointer the books until it proved it could do the work across the hundreds of edge cases her team handles every day. After all, this is the work that keeps Torfs running.

The smallest mistake here can have catastrophic consequences:

1. A day booked wrong means margins, VAT return, and month-end close are all off.

2. A wrong order means Torfs pays for the wrong sizes and the wrong prices.

3. A wrong record means a customer buys a shoe on the web shop at the wrong price.

Jill's team ran Pointer on past months and checked every posted day against what her team had booked. It tied to the cent. To her, this was a strong signal that Pointer could handle messy, real-world work.

Torfs turns the lights off

In 1908, workers built cars by hand, one at a time. Today, most factories run without workers, so no one bothers turning the lights on.

Now, Torfs is turning the lights off, too.

The team gave Pointer the go-ahead, and it now runs several core processes:

1. Daily revenue reconciliation: every day, 10,000+ receipts come in from 150 registers and the web shop. The day's sales are pulled by store and register, checked, and posted to the general ledger, so the accounts clear.

2. Supplier order confirmation: suppliers confirm thousands of order lines a day. Each confirmation is matched against the original purchase order on quantity, price, discount, size, and delivery date, and every mismatch is flagged before the goods arrive.

3. Article creation: 15,000 new products come in a year from about 200 suppliers. Each one is set up in the ERP in every size, with its cost, price, and product details, so it can be bought and sold.

4. Purchase order creation: Once the buyers pick a shoe, the quantities are split across its size range, up to 15 sizes, and the purchase order is raised in the ERP.

Stijn and the team aren't stopping there.

What started with one process - daily revenue reconciliation - quickly grew into these four, with even more processes in the works. And it's not just our deployment team that is setting these up. Torfs' own tech team is working on the same Pointer platform, and on their own they've built two more processes.

Process by process, Pointer is helping Torfs dim the lights.

Torfs gets back to growing

Retail is a cutthroat market. Torfs can't afford to play defense. Every minute and Euro not spent on growth gives ground for their competitors to catch up, and Stijn knows it: "size is a valuable thing there."

So Torfs is expanding. Last month, the company made its first international acquisition, and has its sights on more stores and a bigger webshop.

But every new store, supplier, and product piles more work behind the 3 million shoe boxes. That work used to land on Jill's desk every morning. Now it lands on Pointer, and the business can grow at the speed of their ambition.

And Jill's team? They can look up from the receipts. In her words: "Now that we book the sales with Pointer, we have more time to do interesting things, more time to see what the figures and the data tell us. It adds more value to my job and to the company."

None of that growth changes what makes Torfs, Torfs. Walk into the bright yellow store anywhere from Antwerp to Ghent, and you'll still feel welcome. But somewhere, silently, a back office factory is working with the lights off.

"What excites me most about the future, when I think about what Pointer can automate, is how we can focus on value creation."

– Stijn De Knop, CFO, Torfs

Stijn's bar was A to Z. Not 70%, not "mostly automated." Most teams settle for less, and that's fine. But if your bar is as high as Stijn's, we'd love to talk.