Coaching Spotter

Coaching helps Spotter understand how your organization thinks about data by providing guidance 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 Spotter notes ambiguity after the data model has been optimized.

For detailed guidance on prerequisites and the recommended order of using coaching tools, 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.

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, expecially 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.

Where coaching is stored

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 global and apply to all queries on the data model.

  • Reference questions and business terms are specific to the data model but can be applied specifically at the user level (that is, only the user who created the coaching will have it apply to their searches).

  • Reference questions can be added from a conversation. Business terms can be added directly from the Data workspace.

All coaching influences how Spotter responds to relevant questions on the Model.

Get started

If you’re new to coaching, we recommend this approach:

  1. Optimize your data model and metadata.

  2. Test Spotter with common questions.

  3. Add coaching only where Spotter gets confused.

  4. Start with global rules, then add examples and definitions as needed.

For guidance on choosing the right coaching tool and avoiding common pitfalls, see Coaching best practices.

Was this page helpful?
×