Invoice Capture

Invoice capture is the first step in the accounts payable process: getting the data from a supplier invoice into a system where it can be processed. In a manual AP process, capture means someone reads the invoice and types the relevant information into Xero, MYOB, or a spreadsheet. In an automated AP process, capture means software extracts the data from the invoice document and structures it for the next step.

The quality of invoice capture determines the quality of everything that follows. If the captured data is wrong, the coding will be wrong, the matching will fail unnecessarily, and the approval process will be working from incorrect information. Garbage in, garbage out applies directly to AP: a system that starts with poor data extraction produces unreliable results downstream.

How invoices arrive

Australian businesses receive supplier invoices through several channels. Email is the most common, with PDFs attached. Some suppliers use supplier portals, requiring the business to log in and download invoices manually. A smaller number of suppliers still send paper invoices by post, which require scanning before they can be processed digitally. A growing number of businesses are beginning to receive Peppol eInvoices, which are structured data files that do not require OCR processing at all.

Each channel has a different capture requirement. Email invoices can be forwarded to a capture inbox and processed automatically. Portal invoices require a manual step to download them before capture can begin. Paper invoices require scanning. Peppol eInvoices arrive as structured data and can be processed without any extraction step. Managing multiple channels without losing invoices is itself an operational challenge that capture automation addresses.

OCR and its limitations

Most AP automation software uses optical character recognition to extract data from invoice PDFs. OCR reads the text on the document and attempts to identify which text corresponds to which field: invoice number, date, supplier name, ABN, total amount, GST amount, and individual line items.

OCR works well for invoices with a clear, consistent layout. It struggles with handwritten invoices, poor-quality scans, invoices with complex tables, and invoices from suppliers who change their format regularly. When OCR fails to extract a field with sufficient confidence, the invoice is flagged for human review. The rate at which invoices require human intervention after OCR extraction is called the exception rate, and reducing it is one of the primary goals of capture quality improvement.

Line-item capture versus header capture

There are two levels of invoice capture. Header-level capture extracts the summary fields: supplier, date, invoice number, total, and GST. This is sufficient for simple invoices from suppliers with a single account code. Line-item capture goes further, extracting each individual line on the invoice: description, quantity, unit price, GST treatment, and any job or project reference.

Line-item capture is required for invoices that need to be coded across multiple accounts. A freight forwarder invoice with charges for international freight, fuel surcharge, import duty, and customs clearance needs each line coded separately to the correct account with the correct GST treatment. Header-level capture cannot support this; the entire invoice would be coded to a single account, which produces incorrect management accounts and potentially incorrect BAS figures.

Capture as the foundation of automation

The capture step is the entry point for the entire AP automation process. A business that captures invoices accurately and completely from every channel can then automate coding, validation, matching, and approval. A business that relies on manual data entry for capture has a bottleneck that limits the value of every downstream automation step. Investing in reliable capture is the prerequisite for AP automation to deliver its full benefit.