Transport company (60 employees): invoice 3 days after delivery instead of 12

Example case · anonymized · figures are illustrative

The situation

A transport company in the Rotterdam–The Hague region does distribution in the Benelux. It has 60 employees, 38 of them drivers and 4 in administration, and runs about 350 trips a week for 40 regular clients.

Planning runs in a TMS, accounting in Exact. The drivers use a phone app to photograph the signed CMR or POD. On paper, everything is in order. In practice, there is a twelve-day gap between the photo and the invoice.

The process before Optimi

1. A driver delivers, has the CMR signed, and takes a photo. Sometimes sharp, sometimes at an angle, sometimes a day later.
2. The photos land in a shared folder. An administrative employee opens each photo, reads the sender, recipient, and date, and looks up the trip in the TMS.
3. She links the photo to the trip. If the photo is unreadable or missing, she calls or messages the driver.
4. At the end of the month, she creates a consolidated invoice per client, with all the PODs attached.
5. If a POD is missing, the trip is left off the invoice or the invoice is disputed.

Together, the four administrative employees spent 22 hours a week matching and chasing.

Where it failed

Invoicing ran twelve days behind delivery on average, because everything waited for month-end and for the last missing PODs. Six percent of invoices were disputed, almost always because the proof of delivery was missing or unreadable. Every disputed invoice meant searching again, emailing again, and another two weeks waiting for the money.

The owner-director saw it in his cash flow. A week off the payment time was worth more to him than the hours saved.

What Optimi did

Scan. Two weeks shadowing administration and planning. CMR-to-invoice was the process with the most hours and the biggest financial impact. We also found a second candidate (planning emails), which followed later.

Prototype in 10 business days. We built the prototype for the ten largest clients, good for 60% of the trips. On day 9 we tested it on 1,200 historical CMR photos.

What is in place now:

• An intake that picks up every photo or PDF from the driver app.
• An AI step that reads the CMR fields: sender, recipient, date, reference, and whether there is a signature.
• An automatic link to the trip in the TMS based on those fields. Below the confidence threshold, the photo goes to a queue that administration reviews.
• Every morning, a list of yesterday’s trips without a POD, straight to the planners, so they can speak to the driver the same day.
• Per client per week, a draft invoice in Exact with the PODs attached. Administration approves, Exact sends.

A person stays in the loop for: every uncertain match, every missing signature, and every invoice before it is sent.

Build and Managed. After the prototype, the flow was extended to all clients, and the company chose Managed AI. New CMR forms from new clients are ours to handle; administration does not have to change anything.

The result

Indicative results for this scenario, after one quarter:

• Administration on CMR processing: from 22 to 6 hours a week.
• Invoicing: from an average of 12 to 3 days after delivery.
• Disputed invoices: from 6% to 1.5%.
• DSO: eight days shorter. For this company, that meant structurally more room on the line of credit.

Handwritten CMRs remain hard. About one in ten goes to the queue. That is exactly the point: administration now looks only at the cases where a person is needed.