Homepage
>
Ressources
>
How to extract invoices and PO data from Sage Business Accounting?

How to extract invoices and PO data from Sage Business Accounting?

We explain how to use Sage’s built-in OCR features (AutoEntry) and compare their performance with external OCR solutions.

Author and Co-Founder at Koncile
By 
Jules Ratier
Last updated: 
June 24, 2026
 - 
8 min read

You have two options: use Sage’s native AutoEntry tool, or connect an external OCR through an automation platform like Make. This guide walks you through extracting invoice line items, supplier-specific details, and advanced industry fields.

Option 1 – Use Sage AutoEntry: built-in but limited

What is AutoEntry?

AutoEntry is Sage’s integrated data capture tool, originally developed by a startup acquired by Sage in 2019. It can process various types of documents including invoices, receipts, expense reports, and credit card statements.

From the web or mobile app, it allows you to extract invoice line items and push supplier invoices directly into your accounting system.

The main benefit: native integration with Sage. Uploaded documents are automatically categorized, matched, and synced with your accounting records—minimizing manual input.

Where AutoEntry falls short

Despite offering line-item extraction, many users report accuracy issues—especially with complex tables or supplier-specific formats.

AutoEntry doesn’t allow for custom templates based on prompts or specific instructions. It mainly relies on traditional machine learning algorithms, which are less effective on complex or highly variable documents.

Key user feedback includes:

  • Frequent need for manual corrections after OCR processing
  • Regular outages or service downtime (sometimes up to a week)
  • An initial setup that many find complex or unintuitive

Option 2 – Connect Sage to an external intelligent OCR

For more advanced data capture needs, it’s recommended to use an Intelligent Document Processing (IDP) solution that goes far beyond basic OCR—offering smart document splitting, classification, data cleaning, field matching, and prompt-based automation.

Step 1: Build a workflow from Sage to an external OCR

With a tool like Make, you can create an automated scenario connected to Sage. A typical process looks like this:
• Detect the arrival of new supplier invoices
• Automatically download the files from Sage
• Send the file to an external OCR, such as Koncile OCR

Before choosing a tool, make sure to check the list of apps compatible with Make—not all OCR solutions offer smooth integration with Sage.

Step 2: Create a template for the data you want to extract

You can configure a custom template to capture exactly the fields you need from your invoices.

  • The “General” tab lets you select common fields such as supplier name, date, total amount, etc.
  • The “Table” tab allows you to extract invoice line items, with full control over each column (e.g., product reference, quantity, description).

Each field is enriched with a confidence score, helping you filter out low-quality or unreliable documents.

Step 3: Set up smart document classification

If you handle a wide variety of documents (e.g. energy invoices, transport, service providers), you can create specialized templates for each invoice type.

For standard documents, one comprehensive template may be enough.
For more specific cases, you can configure automatic classification to route each document to the right template.

💡 When in doubt (low confidence score), you can add a human review step on a small percentage of documents—typically less than 1%.

Step 4: Export the data or inject it automatically

The extracted data is available in CSV or Excel format for immediate use.

To take it a step further, you can set up an API connection: data in JSON format is automatically pushed into Sage via webhook—no manual input required.

What can you extract from a supplier invoice?

General Information

Modern OCR tools (enhanced by LLMs) can efficiently extract standard fields such as: Supplier name, invoice date, total amount (incl. tax), invoice number, and more.

Line Items

Extracting line by line remains a challenge—especially with detailed POs. Fields like product references (EAN, SKU, GSIN), quantities, descriptions, and logistical units are often hard to extract reliably and completely. Even with electronic invoices, line-level data can be too limited to produce clean, structured output.

Specific or Industry-Specific Data

Some industries require very precise fields: Transport (Incoterms, freight codes, lot numbers), Energy or Construction. With advanced OCR solutions, it becomes possible to extract these complex fields with high accuracy and consistency.

The agents that automate your documents

Get ahead on automation. See how Koncile can simplify your operations.
Discover Koncile
Discover Koncile
Our latest ARTICLES

Real life insights on document automation

All our ressources
All our ressources
News

Developers launch MacDoc, an illegal fake passport generator

In February 2026, US authorities shut down OnlyFake, a widely used fake document generator. Weeks later, an identical service reopened under the name MacDoc. Its forgeries are getting past the KYC checks of major institutions including PayPal, Airbnb and Coinbase.

Read the article
FEATURE

Document deepfakes: catch the fakes slipping past your checks (2026)

In 2026, document fraud has a new face: the document deepfake, a fake generated entirely by AI. According to Shufti's Identity Fraud Index 2026, it's the fastest-growing fraud category, with an annualized increase estimated at nearly 3,900%. What sets these fakes apart: they're clean on the surface. Here's how to detect the ones still slipping past your checks.

Read the article
Comparative

Top 10 Best Document Fraud Detection Software in 2026

Document fraud has outgrown Photoshop. In 2026, the fastest-growing threat is the AI-generated fake: invoices, bank statements and payslips produced from scratch, pixel-perfect, with clean and consistent metadata. Here is our comparison of the 10 best document fraud detection software platforms every finance and risk team should know, and how to choose the right one.

Read the article
Comparative

5 Best French OCR Solutions to Extract Data from Your Documents

Five French OCR solutions now make it possible to automatically extract data from your invoices, contracts, and accounting documents using optical character recognition, with hosting based in France. Here's our overview.

Read the article
FEATURE

MCP OCR: How Koncile Lets AI Agents Extract Documents Natively

Until recently, connecting an OCR engine to an AI assistant meant writing custom code, managing API calls manually, and building the glue between your language model and your document processing pipeline. With the Model Context Protocol (MCP), that entire layer disappears. At Koncile, we built an MCP OCR server so that Claude, Cursor, or any compatible AI agent can extract, read, and manage documents without a single line of integration code

Read the article