Coaching Spotter
Coaching helps Spotter understand how your organization thinks about data by providing guidance assets that shapes how questions are interpreted and answered.
Rather than enforcing fixed responses, coaching supplies examples, rules, and definitions that Spotter can draw from to reason more accurately-- both within a conversation and across similar questions.
What is coaching in Spotter?
Coaching in Spotter is the process of teaching the system how to interpret intent, apply logic, and resolve ambiguity when answering questions.
Coaching does not store answers or guarantee identical responses. Instead, it provides signals that influence how Spotter reasons about questions over time.
When to use coaching
Coaching is intended as a next step after data model optimization.
Before applying coaching, ensure that your data model has clear column names, accurate descriptions, and appropriate context. In many cases, improving metadata or model structure resolves issues without requiring any coaching.
Coaching should be applied only when genuine ambiguity remains after the data model has been optimized.
For detailed guidance on prerequisites and sequencing, see Coaching best practices.
How coaching improves responses
When applied thoughtfully, coaching helps Spotter:
-
Better interpret ambiguous or incomplete questions.
-
Apply consistent business logic and defaults.
-
Generalize correct reasoning across similar questions.
-
Reduce repeated clarification.
Coaching works best when paired with a well-structured data model and clear metadata, also known as semantics.
Spotter coaching tools
Spotter offers multiple coaching tools, each designed for a different type of guidance.
Data model instructions
Define global rules and defaults that apply broadly across questions on a data model.
Use instructions to establish consistent behavior and resolve recurring ambiguity.
Learn more about Data model instructions.
Reference questions
Provide guiding examples that Spotter can use when reasoning about similar questions.
Reference questions help Spotter learn patterns of intent and logic, especially when paired with clear natural language context.
Learn more about Reference questions.
Business terms
Define rigid, universally true meanings for specific business vocabulary and metrics.
Business terms ensure Spotter uses your organization’s exact definitions rather than generic assumptions.
Learn more about Business terms.
Memory
Memory is a new, AI-generated context layer that works alongside your existing coaching. Where coaching requires you to author and maintain every rule and example manually, Memory is generated automatically from trusted Liveboards you specify to Spotter, or from definitions and corrections shared during a conversation.
Memory is additive. Your existing coaching continues to work exactly as before, and is not affected.
Learn more about Memory.
Where coaching lives
Coaching is applied and managed at the data model level, though some tools can be created from different areas of ThoughtSpot.
-
Data model instructions are managed on the data model, and can alos be modified directly from within a Spotter conversation.
-
Reference questions, business terms, and memory from Liveboards are managed from the Memory sources page in the Data workspace.
-
Memory generated from Liveboards or conversation can be fine-tuned directly in a Spotter conversation-- corrections and new definitions taught mid-conversation are saved as memory and apply to future questions.
All coaching influences how Spotter responds to relevant questions on the Model.
Get started
If you’re new to coaching, we recommend this approach:
-
Optimize your data model and metadata.
-
Test Spotter with common questions.
-
Add relevant Liveboards that contain your business logic on the associated data model to help Spotter understand basic definitions.
-
Verify with similar common questions and correct any learnings that are inaccurate.
-
Add instructions, reference questions, and learning from conversation to fine-tune the experience.
For guidance on choosing the right coaching tool and avoiding common pitfalls, see Coaching best practices.
Additional Spotter coaching resource
How to Coach
ThoughtSpot Spotter
Includes planning, testing, launching, and continuously improving coaching.
View guide
