Import Unstructured Data for Text Analytics

Overview

Medallia Experience Cloud’s advanced Text Analytics capabilities apply topics, themes, sentiment, and more to text phrases in unstructured textual data. Organizations can provide this unstructured data gathered from other sources to Medallia to apply Text Analytics algorithms to their data.

Example Use Cases

• Case notes: Upload notes from contact center agents to discover trends on how agents are responding to specific case types.
• Chat transcripts: Upload chat transcripts from agents conducting live chats to determine if the customer may be a good candidate for an up-selling action.

APIs Used

• Data Import
• Custom Import

📘

Medallia Text Analytics is a paid subscription feature of the Medallia Experience Cloud. Contact your Medallia representative for additional information about how to add it to your subscription.

📘

Data Import API from the Agent Notes Analyzer App

The example below and its endpoint is configured to import customer transactions in JSON format. Depending on your use case and apps installed, you may have different, pre-created APIs or create your own. You can go to the Install Apps page in Admin Suite to see what apps you currently have on your instance or the Medallia Xchange to explore new apps.

Approach

Send data to the Data Import API, targeting the /inbound/v1/active_cc_agent_notes_import endpoint.

The incoming payload and its unstructured text data is processed through Medallia’s ETL layer.

Data is then analyzed through Medallia’s Athena processing engine which applies topics, themes, sentiment, and more to text phrases.

View text analytics data in reports and trigger downstream workflows based on the results of the processing.

📘

Ensure Text Analytics topics are created prior to importing unstructured data

In order to get the most out of unstructured data, create text analytics topics in Medallia Experience Cloud before importing that data.

If importing chat or call logs, see Conversational Transcript Requirements for directions on how to appropriately format your data.

HTTP Request/Response

The below request imports the CASE_NOTES from a call center agent for further processing through Text Analytics:

POST /inbound/v1/active_cc_agent_notes_import HTTP/1.1
Host: example.apis.medallia.com
Authentication: Bearer ${TOKEN}
Content-Type: application/json
Accept: application/json

[
  {
    "CASE_ID": "2210902978",
    "CASE_OPENED_DATETIME": "2020-06-11 12:00:00",
    "CASE_CLOSED_DATETIME": "2020-06-17 12:00:00",
    "CASE_CREATED_BY_ID": "JB0057820",
    "CASE_OWNER_NAME": "Jane Brown",
    "CASE_OWNER_ID": "JD0057820",
    "CASE_STATUS": "Closed",
    "CASE_SUBJECT": "Login Issues",
    "CASE_NOTES": "Customer was unable to access mobile app for two factor authentication. Screened caller to ensure identity and provided password reset instructions. Stayed on the line until the customer was able to log in."
    "CUSTOMER_ID": "22901189"
    "FIRST_NAME": "Scott", 
    "LAST_NAME": "Smith", 
    "EMAIL_ADDRESS": "[email protected]", 
    "PHONE_NUMBER": "555-555-5555", 
  }
]

HTTP/1.1 200 Ok
Content-Type: application/json

{
  "records":1,
  "duplicates":0,
  "rejects":0
}

Note that:

• It accepts JSON, XML, and delimited text.
• Expected body parameters are defined by ETL configurations.

📘

Import APIs Are configuration-dependent

Required and optional parameters in the body of the API call are defined by the configuration of the Data Import APIs. Check your API configurations for exact specification.

Variations

In cases where timeliness is less important (e.g., surveying on long-lived, non-transactional experiences), data can be uploaded to Medallia ETL via SFTP or S3 file transfer.

Conclusion

Analyzing unstructured data is an easy and rich way to glean feedback from your customers, leveraging both solicited feedback (ex: survey comments) and natural communications (ex: chat logs). Medallia Text Analytics uses both topic and sentiment analysis via the Athena processing engine to provide you with those insights. Raw input data can be provided to using one of our Import APIs.