Available in the Pro and Complete subscriptions.
With a custom recognition field at line level, you can automatically recognise values from the individual lines of an invoice. For this to work, it is important that line recognition has been trained, or that the lines are already visible in the recognised data. You can check this by clicking the three dots in the top right of a transaction and opening Recognised data. If no lines appear, you can train line recognition using the Line recognition button in the transaction screen. Refer to the article on training line recognition for more information.
A commonly used example is recognising licence plates per invoice line, which is useful for invoices from leasing companies, petrol stations, or garages where a different vehicle is listed on each line.
This article walks you through the following steps:
Creating a custom recognition field at line level (simple mode or prompt)
Setting up a Financial transaction template with line recognition
Viewing the result in the transaction screen
Step 1 – Creating a custom recognition field at line level
Navigate to Manage > Company (or Organisation) > Custom recognition fields and click the green New custom recognition field button in the top right.
Fill in the fields as follows:
Field | Input |
Name | Give the field a clear name, for example: Licence plate. The name must be unique at company and organisation level. If the name is already in use, a cross will appear. |
Name translations | Only applicable when there are users who work in another language. If filled in, this overrides the Name field, making the translation visible under the recognised data. |
Level | Select Line. The field is then recognised per invoice line. Zenvoices continues searching until all values have been found, which is useful when a field appears multiple times on a document, such as a list of licence plates. |
Please note: Manual line recognition via the Line recognition button in the transaction screen is currently not yet supported. The lines must already have been recognised in the recognised data, or you can train this via manual line recognition training. Refer to the article on line item recognition for more information.
Field | Input |
Data type | Select Text. A licence plate consists of letters and numbers. |
Mode | Select Simple or Advanced. Simple: enter the text that appears literally on the invoice in the Field name field. Zenvoices then automatically searches for the corresponding value. Advanced: write a prompt if the licence plates do not have a fixed field name or column heading on the invoice. |
Make available as transaction variable | Enable this option if you want to use the recognised value in the description of the lines. If this is not needed, you do not need to enable it. |
Tip: Use the variable {licenceplate} in the description field of the transaction line to automatically include the recognised licence plate in the transaction description.
The remaining options are not applicable for this example.
Click Save.
Step 2 – Setting up a Financial transaction template with line recognition
Now that the custom recognition field has been created, set up the Financial transaction template. Set the template to line recognition, as the line type is only supported via line recognition.
Add the licence plate field to the Cost centre field in the template. Then select Match fields and specify which value to match on. This can be the code or the name of the cost centre. It is important that the cost centre corresponds to the recognised field.
If the cost centre has a different name or code than the recognised licence plate, add a lookup table to your prompt so that the system converts the licence plate to the corresponding cost centre instead of returning it literally.
When matching, you can choose between contains, starts with, ends with, or equals. Note the following: if the recognised field has the value 17-FGP-5-B17569-506 and the cost centre has the name 17-FGP-5, then contains checks whether 17-FGP-5 contains the text 17-FGP-5-B17569-506. That is not the case. In that situation, choose starts with.
Refer to the article on setting up a Financial transaction template for more information.
Step 3 – Viewing the result in the transaction screen
After setting up the template, you can link the transaction method at relation level via Manage > Company > Relations > pencil icon. Refer to the corresponding article for more information on setting this up at relation level.
You can also create a transaction method assignment rule to apply the transaction method automatically. Refer to the corresponding article for more information.
Then submit the invoice or re-import it. After processing, the recognised licence plates are visible in the transaction screen under Recognised data. Click the three dots in the top right of the transaction to open this. Click the button with the three dots next to a recognised line to expand the details per line. The recognised licence plate is then visible for each invoice line.
Are the licence plates not showing immediately? Apply the Financial transaction template or change the relation. The recognition will then be run again.
You may encounter three situations:
The field does not appear in the lines: apply the template again. If the field still does not appear, contact Zenvoices at support@zenvoices.com.
The field appears but without a value: the prompt or simple mode is not returning a result. Adjust the custom recognition field and re-import the document to view the result.
The field appears in the lines with a value: the recognition was successful. If the value is incorrect, adjust the custom recognition field as described in step 1.
Frequently asked questions
Can I use the licence plate in an automation rule?
Yes, enable the option Available in transaction automation rules when creating the recognition field. Please note: the field will then be recognised for every transaction.
How do I write a good prompt?
Write an instruction in natural language that describes which information needs to be extracted from the invoice. The clearer and more specific the prompt, the better the result. Read more here.