Connector best practices

Use the following best practices and tips to structure files and configure connectors for optimal performance.

Connectors are integrations that ingest signals — such as audio data, chat, or email transcripts — into Medallia Experience Cloud. They are categorized as:

  • API-based — this type of connectors pull or receive data via API.

  • File-based — these connectors pull or receive data from a flat file, stored in an SFTP or S3 location.

Connector configuration

When creating or editing a connector, the configuration process is split across multiple tabs, each responsible for a different aspect of how data is pulled and processed.

Connector settings

Connector settings control how the connector connects to external storage and reads data.

  1. When configuring file-based connectors:

    1. Use separate folders for metadata and conversation files.

      Store metadata and conversation files in distinct folders so that fetching metadata does not involve heavy conversation files, improving processing performance.

      Sample screen capture of Connector settings tab showing how to configure separate folders for files

    2. Organize metadata by unit.

      • Further divide metadata into sub-folders based on a unit of your choice (such as language or region).

      • Create separate connectors for each sub-folder to parallelize and optimize processing.

        • For example: organize metadata as Metadata/English, Metadata/Spanish, and so on, and configure connectors accordingly.

      Screen capture showing 2 separate connectors: one pointing to the folder with conversations in English and a different one pointing to the folder with conversations in Spanish

    3. Use the metadata file filter input option

      • As an alternative, use a regex filter to select specific metadata files instead of folders. When using this option, make sure you follow a consistent naming convention.

        • For example: if files are named filename_language.csv (such as file1_spanish.csv , file2_english.csv ), use the regex ^.+_english\.csv$ to process only English files.

      Screen capture showing 2 separate connectors: one filters metadata files based on the English language using regex, and the other one that does the same for the Spanish language

Data filters

Use this tool to define filters and transformation on incoming data.

  1. When configuring both, file-based and API-based connectors:

    1. Use Data filters to create formulas that filter files or records based on incoming fields (such as speech file name). This allows for dynamic and flexible filtering for targeted file processing in a connector.

      • For example: filter conversations using the SEARCH formula:

      Custom formula editor canvas

      1. Select a function that performs an action on an incoming field. In this example, we use SEARCH to find a specific string of characters in a text field.

        Screen caption of the formula selector showing the SEARCH function

      2. Choose an incoming field.

        Use the incoming field selector to find and select the right field

      3. Use this field to check if the file name has the "_spanish" substring — following the naming convention defined above — to filter out Spanish-language conversations.

        Use the test tool to see if your formula works

Connector schedule

The Connector schedule tool helps you determine when and how often the data is pulled into Medallia Experience Cloud. Connector pull frequency should be proportional to conversation arrival rate: polling the third-party store at a fixed high frequency (such as every minute) is inefficient when the rate of new conversations is low (for example 1 or 2 per hour).

See below a simple heuristic approach used as an illustrative example:

New conversation frequencySuggested pull frequency
0 to 10 per hourEvery 60 minutes
10 to 20 per hourEvery 30 minutes
20 to 40 per hourEvery 15 minutes
More than 40 per hourEvery 5 minutes

Connector-specific recommendations

Follow these guidelines to optimize connector performance.

ConnectorSuggestion
Genesys CloudObservations indicate that the connector requires approximately 10 minutes to complete a data pull from the Genesys Cloud platform.

As a best practice, the pull frequency configured in the Connector schedule should be set to a value greater than the observed execution time.