Before you begin your implementation

Text Analytics offers Topic starter sets to cut down the time and effort required to implement a program. These provide numerous topics to begin the topic building process and reduce (but do not eliminate) the time required to build topics for your company. Before you begin, evaluate the starter set offering.

Implementing Text Analytics

Complete the following steps to implement Text Analytics:

1. Optionally, if you are going to use one or more Text Analytics starer sets, request installation as described in Request a topic starter set.

   Important: Topic starter sets are sold separately from Text Analytics. Verify that the company purchased a topic starter set before requesting its installation.

   Although topic starter sets are sold separately, the Text Analytics reporting package is not. Any company that purchases Text Analytics receives a Text Analytics reporting package.

2. Navigate to the Reporting > Text Processing > Global Text Analytics Settings > Settings screen, and verify that these settings are properly set (or change the settings if they are not):

   • Marking Surveys Enabled is on.

   • Automatic Processing Enabled is on.

     Note: By default, the Automatic Processing Enabled property is turned off on sandboxes. For more information, see Sandboxes.

3. Navigate to the Reporting > Text Processing > Topic & Theme Tagging > Settings screen, and verify that these settings are properly set (or change the settings if they are not):

   • Topic Alert Blocking Enabled is on.

   • Automatic Tagging Enabled is on.

   • If the company is using one or more languages with sentiment support, SENTIMENT_SUPPORT is selected for the TextAnalyticsVersion property. For more information, see Text Analytics supported languages.

     Note: After you enable SENTIMENT_SUPPORT, you cannot change the option back to TOPICS_ONLY. 

   • Enable MATCH processing is on.

   • Use spellchecking is off.

4. The next steps differ depending on whether you installed a topic starter set:

   • If you installed a topic starter set:

     1. Navigate to the Reporting > Text Processing > Topic & Theme Tagging > Auto Tagging Pools screen.

     2. For MemberSurveyFields, select the appropriate comment fields.

        Important: Text Analytics reporting packages are designed to report on customer comments. If you are using a Text Analytics reporting package, do not select case management fields unless the company intends for the text in those fields to be treated as customer verbatims.

     3. For MemberUnitGroups, select the appropriate unit groups.

   • If you did not install a topic starter set:

     1. Create one or more tag pools, as described in Creating a tag pool.

        Create a tag pool to define the pool of comments to which each list of topics is applied. The process for creating a tag pool also creates a new topic set with the same name as the tag pool.

        Restriction: Do not use K-fields and A-fields in tag pools as they may cause performance issues. Use only Q-fields and E-fields.
        Important: Text Analytics reporting packages are designed to report on customer comments. If you are using a Text Analytics reporting package, do not select case management fields unless the company intends for the text in those fields to be treated as customer verbatims.
     2. Add user features for terms that would otherwise be ignored by the parser, as described in User features.
5. If you want to process non-English languages, configure the languages to process.

   • Configure machine translation, as described in Translating comments. Machine translation is the use of software to translate text. Experience Cloud uses SYSTRAN, Amazon Translate, and Google Translate for machine translations. Users with certain permissions can edit comments on the Responses Form report that have been machine translated. Enable this feature if your company requests it.

   • Select topic building languages, as described in Global Text Analytics Settings. If you are you planning to build topics in a language other than English, verify that you or someone working on the project speaks the language, can build topics in it, and has received Text Analytics training.

6. Configure sentiment, as described in Configuring sentiment.

7. Configure Text Analytics for Social Feedback data (if applicable).

   For customers that have purchased and implemented Medallia Social Feedback, Text Analytics can process comments from review websites in addition to survey comments.

8. Initiate text processing, as described in Text Analytics processing and initialization.

   You will not see topic data in your reports until you complete this step. For more information, see Text Analytics processing.

   Note: When processing for the first time, select Reprocess all records as the Processing option.

9. The next steps differ depending on whether you installed a topic starter set:

   • If you installed a topic starter set, customize the topics.
     1. Create and modify the rules for each topic. Your goal should be to:

        • Increase the number of topic matches.

          The topics in the package are optimized for accuracy over coverage, so you will need to add additional keywords and logic to ensure that most of the relevant feedback is captured.

        • Customize the topic for the company.

          For example, you might add the name of the company's loyalty program to a topic's rules.

        • Remove any topics that are not applicable to the company.

          For example, the retail topic starter set includes topics about fitting rooms. If the company does not have fitting rooms in their stores, you should remove the fitting room topics.

        Coverage analysis is one method you can use to boost coverage and estimate the percentage of comments that are matched to at least one topic. For more information, see Test rule-based topics.

     2. Optionally, add additional topics to the topic set.

        Show company stakeholders the standard topics and work with them to decide whether additional topics are necessary.

   • If you did not install a topic starter set, build your topics. While you complete this task, work with your Medallia expert to track your work.

   Tip: For guidance on planning, creating, and managing rules and topics, see the following topics and sections:

   • Plan topics — Provides guidance for creating, refining, and finalizing your topic list.

   • Topics screen — Explains how to create a topic, create topic rules, assess captured comments, and activate a topic. This topic also explains how to assess topic accuracy, read relevant comments, discover new keywords, and when necessary, remove comments associated with automated topics.

   • Rule components, How rules are applied, and Word matching sections in Standard topics — Define rules, rule components, syntax guidelines, rule groups, word groups, and user features.

   • Test rule-based topics — Work with your Medallia expert to test your topic rules and document your results.

10. Run theme discovery, as described in Configure Theme Explorer.

11. Perform one or more of the following steps based on which Text Analytics reporting package is installed on your instance.

    • If the generic reporting package is installed, complete the configuration steps listed in Configuring the Generic Text Analytics reporting app.

    • If any Text Analytics reporting package is installed, review the the overview and health check dashboards to determine whether they need to be updated to include Text Analytics data. See Report modules in Text Analytics reporting apps for more information.

12. Optionally, enable sentiment overrides, as described in Sentiment overrides (opt-in feature). 

    Sentiment overrides allow companies to edit the sentiment of individual sentences.

13. Optionally, Configure topic alerts.

    Topic alerts trigger a response when the text in a survey record matches a Text Analytics topic.

14. If your company is using Action Intelligence powered by Medallia AI features, configure those features as described in Action Intelligence.

15. After you've tested the configuration, launch Text Analytics:

    1. For companies that have purchased Self-Service Text Analytics, see Providing access to self-service Text Analytics to provide access to the Topics screen.

    2. Provide all relevant roles with the appropriate Text Analytics permissions, listed in Reporting permissions and Administrative permissions.

    3. Provide all relevant roles with access to the Text Analytics reporting dashboards.

       • Corporate roles with access to a large amount of data typically get access to the full suite of Text Analytics dashboards.

       • In general, the less data a role has access to, the fewer Text Analyticsr reports the role needs. Frontline users typically do not have access to Text Analytics because Text Analytics works best with a large amount of data, and frontline users typically do not have access to enough of that data. When frontline users do have access to Text Analytics, it is limited access; perhaps only to Responses and volume-based reports (as opposed to reports with impact scores or sentiment).

       • Crosstab reports are applicable only to insights teams or power users

       • Access to Text Analytics is usually rolled out slowly. Corporate roles typically get access first. Oftentimes, it makes sense to let a few key stakeholders acclimate to the system before rolling it out to additional roles.

    4. Verify that each role with access to Text Analytics reporting can view topics. See Manage topic visibility in Medallia Setup for more information.

    5. Activate any topic alerts (if applicable). For more information, see Configure topic alerts.