Skip to content
All field notes
Applications25 August 20266 min read

Automating return order processing: what works

Automating return order processing sounds straightforward, but in practice return orders are one of the most stubborn document types in back-office administration. They arrive as standalone emails, PDFs, Word files, or Excel exports, each customer using a different format and rarely including a clear reference to the original order. Below we explain why that is, what it concretely takes to automate it, and when you are better off not automating.

By Yeslin Beljaars

Why are return orders harder than regular orders?

A purchase order or sales order typically comes from a single sender with a fixed format, an order number, and a structured line layout. A return order is the opposite: the customer sends back whatever they want to return, in whatever way suits them. That creates a few concrete problems. First, free-text fields: the reason for return appears as running text in the email or in a description field with no fixed options. Second, missing references: not every customer includes the original order number or item codes. Third, inconsistent formats: one customer sends an Excel sheet with column names that differ slightly every time, another sends a PDF with a handwritten table. On top of all that, return orders are often time-sensitive: the customer wants to know whether the return has been received and when to expect their credit note.

What makes automating return order processing concretely difficult?

Classic template-based processing does not work for return orders, because there is too much variation per customer in layout and field names. You cannot build a template for a hundred customers each using their own format. Another pain point is the missing reference: if a customer does not include an order number, someone has to search manually in the ERP or WMS to identify which original shipment is involved. That costs time and introduces errors. The return reason also plays a role: for credit note processing, inventory management, and quality reporting you want that reason captured in a structured way, but in most cases it is written as free text. Converting that free text into a fixed code (damaged, wrong product, over-delivered) is an interpretation step that is not trivial.

How does dottle process return orders automatically?

dottle reads incoming return orders without requiring you to set up a template per customer. The system recognises the document structure, extracts item codes, quantities, reasons, and references to original orders, and converts them into structured data that feeds directly into the right fields in your ERP, WMS, or accounting package. If a reference is missing, dottle flags it as an exception so a team member can look it up specifically, rather than retyping the entire document by hand. Return reason text is recognised and mapped to predefined codes that you configure. A person approves the processing before the system creates the credit note or updates inventory. That is intentional: dottle handles the reading and data entry, the decision stays with the user. Varying formats per customer are not an obstacle: dottle works even when the layout differs from one shipment to the next, with no need to set up a new template each time.

When does automating return order processing not pay off?

Honesty about the limits is warranted here. If you process dozens of return orders per month, the payback period of an automation solution is long. Setup time and ongoing maintenance will likely outweigh the manual effort saved. The same applies when the return process is full of exceptions that each require a separate conversation: partial returns with damage claims, disputed quantities, or customers who change their request mid-process. In those cases, the real bottleneck is not entering the document but the decision-making around it. Automating the document processing then helps only to a limited degree. Finally, if source information is structurally absent, no tool can fix that. If customers do not include item codes or order numbers and that does not change after communication, manual lookup work will always remain.

How does return order processing connect to your existing systems?

dottle delivers the structured return data into the system where the work happens: SAP, Dynamics 365 Business Central, Exact, AFAS Profit, Unit4, or a WMS. In Business Central, you can already automatically create a corresponding purchase credit note from a purchase return order; dottle then fills the fields needed to initiate that step. The credit note processing itself is the next link, handled by your accounting package. The integration path is part of the setup; dottle goes live in two weeks, with no setup fee and no lengthy implementation project.

Seeing this in your own document flow?

Book a demo on your own documents

Frequently asked questions

How do I automatically process return orders in my ERP?

A document AI tool like dottle reads the incoming return order, extracts item codes, quantities, and reasons, and delivers them as structured data into your ERP. A team member approves the processing before the system creates the credit note or updates inventory. This avoids manual retyping and errors caused by missing references.

What is the difference between a return order and a credit note?

A return order is the customer's request to send goods back, including the reason and quantities. The credit note is the financial document you issue as the seller once the return has been processed and approved. Automation focuses on reading the return order; the credit note follows as a separate step in the accounting or ERP system.

Does automated return processing work when customers do not include an order number?

Partially. A document AI tool can still extract item codes, quantities, and the return reason, but if an order number is missing, a team member has to look up the original order manually. A good system flags this as an exception, so the team member can take targeted action rather than retyping the entire document.

When is it not worth automating return orders?

At low return volumes (dozens per month), the payback period is often too long. When the return process involves many exceptions and decision points handled by phone or email, document automation only solves a small part of the problem. And if customers structurally omit item codes or references, manual lookup work will always be necessary regardless.