Create Order and Create Return: Setup in Mediabriz

Modified on Thu, 15 Oct at 5:32 PM

Create Order, Create Return, or both in Mediabriz. Each action uses AI to prepare a draft from the customer conversation; the agent reviews and approves it before it is sent to your system. Use the same setup steps below, choosing the appropriate action type.


1. Prepare the connection details

Make sure you've completed the previous steps in the Actions Guide.

Have these details ready:

SettingWhat you need
Webhook URLThe HTTPS address that will receive approved orders or returns.
Webhook secretThe receiving system’s signing secret. It is required even if Authentication is None.
AuthenticationWhich option to select: None, Bearer token or Custom header. For Custom header, use its configured name.
TimeoutHow long Mediabriz should wait for a response, from 3 to 20 seconds.

Enter these values in step 4. Identify any extra information your receiving system needs, such as a PO number or requested return outcome.


2. Add the action in Mediabriz

Your account needs integration:cloudBriz:view and integrations:integration:view.

  1. Open Apps → Mediabriz → Conversation Actions → Add action.
  2. Under Action type, choose Create Order or Create Return. You cannot change the type after saving. If Create Order or Create Return is missing, contact Mediabriz support to enable it.
  3. Enter a Display label in your agents' language, such as Sales Order or Product Return.


3. Add business context (optional)

Business context helps AI understand product names, units and common terms. Write up to 300 characters in your customers’ language. Rules about which orders or returns are allowed must be checked in your receiving system.

ActionExample
Create OrderWholesale dairy. Customers order by crate; “milk” means 3% unless stated; orders often arrive as a bare product list.
Create ReturnWholesale dairy. Returns are usually damaged or expired crates reported with a photo; customers quote the delivery note number.

Products outside the context may still be extracted and marked Needs attention for agent review.


Optional: add extra fields
If your system needs a PO number, notes or requested return outcome, follow Configure Custom Fields, then return here to connect and save the action. Use the keys expected by your integration. AI does not fill in custom fields.


4. Connect, choose divisions and save disabled

Enter the Webhook URL, Authentication, Timeout (seconds) and Webhook secret prepared in step 1. If using Custom header, enter the agreed header name. The secret is stored encrypted and is not displayed again; enter a new value only to replace it.

Under Divisions, select the divisions whose agents should see this action. Leaving all unchecked includes every division. Select Create as disabled while setting up and testing. For an existing action, use Save changes after editing.


5. Prepare and upload the product list

The product catalog is a CSV file containing the products this action should recognize, including each product's SKU and product name. Mediabriz uses this list to match products mentioned in conversations to your organization's products.

Follow Product Catalog: Prepare and Upload to prepare the CSV, upload it to this action and resolve any import errors. Then return here to test and enable the action.

A required catalog must be uploaded before agents can see the action.


6. Test the complete connection

Test each configured action separately after the receiving system and catalog are ready.

First: send a sample

  1. In Mediabriz → Conversation Actions, review Sample webhook payload. Copy or download the sample JSON if needed.
  2. Select Test webhook. It works while the action is disabled and uses the same address checks, signature, authentication and timeout as a real delivery.
  3. Check the displayed status code and success, referenceId and message, or the failure reason. Check the receiving logs to confirm signature validation and processing.

The sample uses fictional product information and metadata.source: mediabriz-test. It matches the configured action. Return samples include reason, reasonText, reference and pickupDate.

Custom fields in the sample

The test does not read a real conversation. Check the sample values before testing:

  • Fixed or hidden fields set here: the configured value.
  • Editable fields set here: the default, or SAMPLE- plus the key, such as SAMPLE-case_number.
  • Dropdowns: the default option key, or the first option’s key.
  • Genesys participant-data fields: the fallback, or SAMPLE- plus the key.

Then: submit a real order or return

  1. Choose the divisions for testing and enable the saved action with its top switch.
  2. Follow the agent instructions to create an order or create a return. Review and approve the draft.
  3. Confirm that your receiving system creates the intended record for the correct customer. Check products, quantities, dates and custom-field values. For returns, also check the reason, reason text, reference and pickup date.
  4. Check the success response and returned reference. Confirm that an already processed eventId is not processed again.
  5. After the checks succeed, make the action available to the intended divisions.

Retry check: an unchanged retry keeps its eventId. Changing products, customer or date fields, or custom-field values creates a new ID. Treat changed submissions separately.

A sample test alone does not prove real submissions are processed correctly. 
If a test fails, inspect the displayed error and receiving logs. Mediabriz does not retry automatically.


7. Make the action available

After the tests succeed, confirm the intended divisions and leave the action enabled. The switch takes effect immediately. Agents with Genesys Cloud already open should reload it.


To pause or manage actions, see Manage Actions.

Was this article helpful?

That’s Great!

Thank you for your feedback

Sorry! We couldn't be helpful

Thank you for your feedback

Let us know how can we improve this article!

Select at least one of the reasons
CAPTCHA verification is required.

Feedback sent

We appreciate your effort and will try to fix the article