A year ago, in the Las Atalayas industrial estate in Alicante, a distribution company with eight employees called us because their administrative team was on the verge of collapse. Four people spent virtually their entire day on repetitive tasks: keying invoices into the ERP, replying to supplier emails, transferring WhatsApp orders into Excel sheets, and reconciling delivery notes with invoices. When someone was off sick, chaos took over.
Today, twelve months later, that same company has eliminated 70% of its administrative workload. They didn’t fire anyone: they redirected the team to commercial and customer service tasks. The investment was recovered in six weeks. This is a success story of administrative automation in a distribution company and, above all, a useful story for any SMB that recognises its own day-to-day in these lines.
We’re going to walk through exactly what we did, what tools we used, and what real numbers were achieved. No marketing, no empty promises.
The starting point: three clear bottlenecks
Before touching anything, we spent two days at their offices observing the actual workflow. We identified three time sinks that recur in almost every distribution company:
- Supplier invoices in PDF: between 60 and 90 per day. One person opened them one by one, copying amounts, descriptions, and due dates into the ERP. Three and a half hours daily, with frequent transcription errors that later had to be corrected.
- Orders via WhatsApp and email: customers ordered through a thousand channels. The salesperson forwarded them to admin, who retyped the order into the ERP. Lost orders, miscopied lines, frustrated customers.
- Unmanaged inbox: the info@ mailbox piled up enquiries, complaints, and delivery note requests all jumbled together. Nobody knew what had been answered and what hadn’t.
The manager summed it up well: “I’ve got four people doing the work of two, but none of them have time for what really matters.”
The solution: three automated flows, a simple architecture
We designed three independent automations sharing the same infrastructure. No replacement of the existing ERP — the client had been using it for fifteen years and it worked. We built on top.
1. AI-powered OCR for supplier invoices
We set up a dedicated mailbox (invoices@). When a PDF arrives, an n8n flow picks it up, runs it through an OCR service with a vision model, extracts the key fields (issuer VAT number, invoice number, date, base, VAT, total, due date, line items), and inserts them straight into the ERP via its API. If OCR confidence drops below 95%, the invoice is moved to a manual review tray with the fields pre-filled.
Result: from 3.5 hours/day to 25 minutes of review.
2. Centralised WhatsApp and email orders
We connected the WhatsApp Business API and the orders email account to a single flow. An AI model interprets the message (“send me 20 of the usual boxes”), locates the customer by phone number or domain, retrieves their order history, drafts the order, and sends it to the salesperson for one-click confirmation. The validated order is automatically inserted into the ERP.
Result: 0 lost orders in the past six months. Average processing time per order: from 7 minutes to 40 seconds.
3. Automatic classification of the general inbox
Everything that lands in info@ goes through an AI classifier that tags the email (invoice, complaint, sales enquiry, delivery note request, other), assigns it to the right person, and — for simple cases like duplicate delivery note requests — replies directly with the document attached, pulled straight from the ERP.
Result: the info@ inbox went from 200 unanswered emails to an average of 4 by the end of the day.
The real numbers after one year
These are the metrics the client shared with us after twelve months, measured and verifiable:
- Manual administrative work: a 70% reduction in total hours spent.
- Data entry errors: from 8-12 incidents detected per week to 0-1.
- Average customer response time: from 6 hours to 35 minutes.
- Orders processed per day: capacity tripled without hiring anyone.
- ROI: the investment was recouped in 6 weeks. From there on, it’s all net savings.
And the figure that’s harder to quantify but most important: the admin team stopped working overtime, and two of its members moved on to handle customer service and commercial follow-up — areas no one had been able to cover before.
When does this kind of automation make sense?
Not every company needs this level of automation. In our experience, this approach fits when at least two of these criteria apply:
- Measurable repetitive volume: more than 30-40 documents arriving at the company every day (invoices, delivery notes, orders).
- Saturated administrative team: staff working overtime or commercial tasks left unattended for lack of time.
- A working ERP with manual tasks around it: the central system is fine; what’s failing is what happens around it.
- Recurring transcription errors that generate complaints or complex reconciliations.
If you tick two or more of these boxes, there’s a strong chance a similar project would make economic sense for your company. Initial investment usually sits between €4,000 and €12,000 depending on complexity, and ROIs under 3 months are the norm, not the exception.
Conclusion
This success story of administrative automation in a distribution company illustrates something we repeat on every project: it’s not about technology, it’s about processes. AI, OCR, and n8n are tools; what changes the outcome is properly analysing the workflow, identifying where time is genuinely lost, and building something proportional to the problem. No over-engineering, no replacing what already works, and measuring results from day one.
If your administrative team is overwhelmed and you recognise some of the problems we’ve described, your company can probably save many more hours than you imagine.