Receipt Automatching

Receipts and invoices need to be attached to transactions for proper accounting and reporting of expenses. One way to do this is by adding each receipt manually to the respective transaction. Either via web app or via API call.

This can be a bit tedious for customers with lots of receipts a given month or day. We therefore offer a second way of adding receipts, invoice and alike to transactions. The receipt automatching is a process that takes files (pdf or images) or metadata about a receipt as an input and tries to find the corresponding credit card transaction within the Pliant system.

📘

This feature is available since API version 2.1.0!

The receipt automatching algorithm uses either the provided metadata or extracts data from the given file itself via an OCR process. It then uses this data to find a matching transaction. It mainly uses data like the amount, the date, the transaction reference or similar references and also merchant data.

You can use this feature with the following set of endpoints:

1. Hand in a file to get automatched and attached to a transaction
   1. Note: this endpoint is asynchronous, which means you get back an automatchingTaskId which you can then use to check the status of the automatching process, or you can wait until the callback is fired, which informs you about the finished task.
   2. A call to this endpoint could look like it is described in this recipe.
2. Hand in metadata to get back just a transactionId
   1. Note: this endpoint is asynchronous, which means you get back an automatchingTaskId which you can then use to check the status of the automatching process, or you can wait until the callback is fired, which informs you about the finished task.
3. Check the status of a automatching task
   1. By providing an automatchingTaskId you can query the status of this specific task.
4. Get notified about finished automatching tasks
   1. Once an automatching task finishes (successfully or not), you will get notified via a separate callback, as long as you have subscribed to this callback.

Please note, that the automatching algorithm and the OCR process are heuristics! Which means they do not have a 100% accuracy. The current accuracy of the overall automatching process is around 95%. The more data is provided or extracted via OCR, the better the matching result.

🏷️

This feature has to be paid separately!

Each created automatching task is billed with a separate fee, reported monthly to the API consumer.