Technical wholesaler (45 employees): 33 hours a week returned to sales support

Example case · anonymized · figures are illustrative

The situation

A technical wholesaler of fasteners in South Holland has 45 employees, six of them in sales support. About 180 orders a day. Some come in through the online store and EDI, but 60% still arrive by email: a PDF from the client’s purchasing system, an Excel list, or just a few lines in the email itself.

The business is doing well. The clients are loyal, the margins are healthy, the warehouse is in order. The ERP is Exact, and it works fine. The problem is not in the systems but in the part in between.

The process before Optimi

This is how every morning went:

1. A sales support employee opens the shared inbox and sorts the orders by client.
2. For each order, he opens the attachment, looks up the client in Exact, and creates a new sales order.
3. Every order line is retyped: item number, quantity, requested delivery date. For a client that uses its own item codes, he first has to translate them through an old conversion list in Excel.
4. After entering the order, he sends a confirmation by hand.
5. On Monday there are 300 orders, and everything slides into the afternoon.

Three employees together spent about 42 hours a week on this. That is more than one full-time employee doing nothing but retyping what the client had already typed.

Where it failed

During the scan we counted it. Retyping did not just cost time, it was also the main source of errors: 2.6% of order lines had a typo in the item number or quantity. Every error meant a wrong delivery, a return, and an unhappy client. On top of that, an order waited an average of three hours for a confirmation, and longer on Mondays. Clients called to ask whether their order had arrived, and that call cost time again.

So the bottleneck sat in one place: between the inbox and Exact.

What Optimi did

Scan. In two weeks we mapped the sales support processes and measured the hours per step. Order entry was at the top of the list by a wide margin: many hours, lots of repetition, a structured source.

Prototype in 10 business days. We built a working prototype for one client group of 15 regular clients. Days 1 and 2 went on scope and access; days 3 to 8 on building; on day 9 we tested on 400 historical orders; on day 10 it went live for that group.

What is in place now, in plain language:

• An inbox listener in Microsoft 365 that picks up every incoming order.
• An AI step that pulls the order lines out of the PDF, Excel, or email text, whatever the layout.
• An integration with Exact that matches the client and the item numbers, including the conversion list for client-specific codes.
• A review screen for sales support. Green: everything recognized, the order goes through. Orange: an unclear line, someone takes a look.
• After approval, the order is created in Exact and the confirmation goes to the client automatically.

A person stays in the loop for: every orange order, every new client, and every order line with an unusual price.

Build. After the prototype, we extended the solution through workflow automation to all email clients, with logging and error messages.

Managed. The wholesaler then chose Managed AI: we watch the flow, catch new attachment formats, and send a plain-language report every month.

The result

The figures below are indicative, based on this scenario, after three months live.

• Sales support on order entry: from 42 to 9 hours a week. The remaining hours go to checking orange orders and exceptions.
• From receipt to confirmation: from an average of 3 hours to 12 minutes. On Mondays too.
• Line errors: from 2.6% to 0.4%. The remaining errors were in the source, on the client’s side.
• 78% of email orders go through without manual correction.

More important than the numbers: sales support now calls clients to sell, not to ask whether the order is right.