What is the difference between sorting, indexing and routing inbound mail?

These three steps form the processing pipeline. Sorting means placing each item in a category (invoice, registered letter, HR...). Indexing means extracting its metadata: sender, type, date, amount, reference, so it can be found and acted upon. Routing means automatically sending it to the right person and tool based on rules. Well-sorted but un-indexed mail stays hard to use; indexed but un-routed mail sleeps in an archive. The three are complementary and inseparable.

Which metadata should be extracted from a supplier invoice?

For a supplier invoice, the key metadata are: the supplier name and ideally its company registration number, the invoice number, the issue date, the due date, the net amount, the VAT amount and the gross amount, plus a matching reference (purchase order or customer case). These fields enable accounting pre-matching, due-date tracking and precise search. OCR extracts them automatically on most structured invoices, with human validation on ambiguous cases.

How do you ensure an incoming registered letter is processed on time?

The pipeline must isolate registered items at the sorting stage based on the bpost label, place them at the top of the queue and trigger an immediate notification to the executive or legal team, with an SLA to process them the same day (for example within two working hours). The legal deadline runs from the date of receipt, not the date of reading, which makes any latency dangerous. A timestamped log keeps proof of receipt and handling, useful in the event of a later dispute.

Can mail be routed automatically to accounting software or a CRM?

Yes, and that is precisely the goal of a mature pipeline. A platform with an API automatically pushes each document and its metadata to your accounting software, your CRM or the payroll office's space, with no re-keying. Supplier invoices go to accounting pre-filled, customer mail attaches to the existing CRM record, HR documents join a restricted-access space. Integration via API removes double entry and keeps data reliable end to end.

Does automatic mail processing comply with GDPR?

Yes, provided a few principles are respected. Inbound mail often contains sensitive personal data: pay slips, medical correspondence, disputes. The platform's hosting must stay in Belgium or the European Union, access to documents must follow the least-privilege principle via roles, and retention periods must respect the regulations, notably the 7 years for accounting records. This processing activity must be recorded in the company's data processing register.

Inbound mail processing: sorting and routing

Receiving mail is only half the job

Many Belgian businesses believe their mail problem is solved once they have a scanner or a digital mailbox. In reality, reception and digitization are only the first half of the job. The second — and most profitable — half is inbound mail processing: sorting, filing, indexing and routing every item to the right person and the right business tool, on time.

A scanned PDF that lands in a shared folder with no metadata and no recipient is barely better than an envelope left on a desk. It is the processing pipeline that turns a stream of paper into usable information. This guide describes the concrete operational rules, from sorting to accounting integration, and sits within our complete guide to business mail management.

Sorting and classification: defining your categories

Mail sorting starts with a clear taxonomy. Without agreed categories, everyone files their own way and the archive becomes unmanageable. A typical Belgian SME works well with seven to ten categories:

Supplier invoices — to accounting.
Customer mail — quotes, orders, complaints.
Administrative and tax — FPS Finance, VAT, social security.
Legal — formal notices, summonses, contracts.
Bank and insurance — statements, endorsements, claims.
HR and social — payroll office, contracts, sick leave.
Registered mail — absolute top priority (see below).
Informational and ads — batched, low priority.

Classification can be manual (the operator picks the category at scan time) or automatic: an engine analyses the OCR text and sender to suggest a category. On recurring mail — same suppliers, same authorities — automation quickly reaches 80 to 90 % reliability. To build this chain, our article on hardware, OCR and scanning software details the technical building blocks.

Indexing: extracting the useful metadata

Filing is not enough: you must index. Indexing means extracting, for each item, a set of metadata that makes the document findable and actionable:

Sender (name, and possibly company registration number).
Document type (invoice, contract, notice, formal notice).
Document date and date of receipt.
Amount and due date for invoices.
Reference (invoice number, case, customer).

That metadata is captured once, on entry, and follows the document for its whole life. It is what lets you search "all invoices over €1,000 due this week" instead of digging through a folder. Well-indexed mail is found in seconds and faithfully reproduced across the 7 years of legal retention for accounting records in Belgium. That standard matches the one in legally valid dematerialization and archiving.

Routing rules to people and tools

Routing means automatically sending the right document to the right recipient based on rules — not on the sorter's mood. A few concrete examples:

Supplier invoice → accounting: filed in the accounting software, notification to the purchasing lead.
Formal notice or registered letter → executive and legal, as priority, with an immediate alert.
Customer mail → CRM, attached to the existing customer record.
HR document → payroll office, in a restricted-access space.

A routing rule combines a condition (category, sender, keyword) and an action (file, notify, assign). These are the same principles detailed in our guide to automating office mail management. The key is to document those rules so they survive the departure of the person who knows them.

Prioritising registered mail and legal deadlines

Not all categories are equal. An incoming registered letter almost always carries a legal deadline that runs from receipt: formal notice, tax appeal, notice period, service of process. Letting it sit 48 hours can be very costly.

The pipeline must therefore isolate registered items at the sorting stage, surface them at the top of the queue, and trigger an immediate notification to the executive. A formal notice, for instance, opens a response deadline whose start point is receipt, not reading — a principle recalled in our formal notice template and procedure in Belgium. Removing the human factor from sorting on this one category eliminates the costliest incident class.

SLAs and notifications: pacing the processing

A serious pipeline sets SLAs (processing times) per category:

Registered letter: processed and notified same day, within 2 working hours.
Supplier invoice: indexed and routed within 24 hours.
Standard mail: within 48 hours.
Advertising: weekly batch.

Each step triggers a notification to the right actor, and the system keeps a timestamped trail of who received what and when. That log serves both internal management and evidence in the event of a dispute. Without a written SLA, "urgent" means nothing; with one, every item has a clock.

Integrating with accounting and CRM (including via API)

Processing reaches its full value when it feeds your tools directly instead of creating a parallel archive. Concretely:

Supplier invoices flow to the accounting software in the right format, pre-filled with the extracted metadata.
Customer mail attaches to the matching CRM record.
HR documents join the payroll office's space.

A platform with an API automatically pushes each document and its metadata into your system, with no re-keying. That is the bridge between the mail processing pipeline and your existing business processes. Our inbound mail scanning workflow for SMEs shows how these flows connect end to end.

The most common mistakes

Sorting without indexing. Filing in the right folder is not enough if the document has no searchable metadata.
Routing to people, not roles. "Send to Sophie" breaks as soon as Sophie changes job. Route to a role or a functional inbox.
Treating registered mail like the rest. The costliest mistake: a missed legal deadline.
Neglecting GDPR. Mail contains sensitive data; access must follow least-privilege and hosting stay in Belgium/EU.

Summary

Inbound mail processing is a pipeline: sorting, indexing, routing, notifications, integration. Each link has its own rules and deadlines. An SME that formalises its taxonomy, indexes systematically, prioritises registered mail and connects its tools via API turns a daily chore into a reliable, traceable process. To frame your processing pipeline, contact the Bjet24 team.