The Power of Natural Language with Amazon QuickSight Q

Overview

This blog post dives into enabling business users to ask questions about their data using natural language within Amazon QuickSight. Amazon QuickSight Q, a powerful natural language query (NLQ) functionality, empowers users to gain insights from data using everyday business terms, eliminating the need for complex queries or reliance on data analysts.

What is Amazon QuickSight Q and How Does it Work?

This allows anyone in an organization to ask questions using their terminology, fostering data accessibility and exploration.

Here’s the key workflow:

1. Topics: Create collections (topics) of one or more datasets representing specific subject areas relevant to your business users’ inquiries. These topics act as the foundation for Q to understand the context of your data.
2. Data Preparation: Amazon QuickSight Q automates much of the data preparation for your topics. However, you must provide some business-specific context to optimize Q’s understanding.
3. User Interaction: Business users can access Q from the QuickSight console or embedded within your website or application. They can ask questions in natural language and receive answers through visualizations or data tables.
4. Feedback Loop: Continuously monitor usage data and gather user feedback to refine your topics and ensure Q delivers the desired results.

Getting Started with QuickSight Q: Best Practices

This section outlines key best practices for building effective NLQ interfaces using QuickSight Q:

1. Start Small: Begin with a well-defined use case – a set of real-world questions frequently asked by business users. Focus on a limited number of fields to ensure initial success and build user confidence.
2. Leverage Automatic Topic Creation: Amazon QuickSight Q offers one-click topic creation from existing analyses. This automatically selects relevant columns based on their usage in your analysis, including any existing calculated fields.
3. Enriching the Context for Q: While Q understands basic English, you must be guided to comprehend your unique business terminology. Here’s how to provide context:
   • Synonyms: Add synonyms to map business-specific terms (e.g., “gross sales” or “amortized revenue”) to the corresponding data field names.
   • Value Synonyms: Use value synonyms to connect how users might refer to specific data values (e.g., “Freshmen” for “First Years”).
   • Semantic Types: Assign semantic types (e.g., Location, Person) to fields to help Q understand user questions related to “who,” “where,” “when,” and “how many.”
   • Default Aggregations: Set default aggregations for measure values to ensure accurate and meaningful data summaries. For instance, percentages or medians shouldn’t be summed.
   • Primary Date Field: If your data includes multiple date fields, choose a primary date field for answering time-based questions like “when” or “year-over-year.”
   • Named Filters: Define named filters to allow users to filter data using specific terms or phrases (e.g., “failing” for test scores below 70%).
   • Named Entities: Create named entities to return sets of fields as table visuals when users ask for specific details. This provides additional context beyond just a KPI.
4. Driving NLQ Adoption:
   • Support: Provide resources like tutorial videos or newsletters to educate users about Q’s capabilities. Establish a communication channel (Slack, Teams) for users to ask questions and request enhancements.
   • Feedback Loop: Monitor usage data and conduct user interviews to identify areas for improvement. Analyze unanswerable questions to refine your topics and address user needs.

Conclusion

By following these best practices and providing the context for Q, you can empower your business users to use natural language to explore and gain insights from their data independently. This fosters data-driven decision-making across your organization.

Drop a query if you have any questions regarding Amazon QuickSight Q and we will get back to you quickly.

About CloudThat