What-If Analysis Skill

This skill lets the AI Assistant compare different business assumptions directly from chat. Use this skill when you want to model alternative scenarios, compare their impact in one view, and optionally receive a recommendation based on the results. The skill builds on metric creation and creates scenario metrics in MAQL by applying arithmetic changes such as constants or multipliers to a base metric.

How It Works

The What-If Analysis skill compares alternative scenarios by creating scenario metrics and visualizing them together.

When activated, the assistant:

1. identifies the base metric and the scenario assumptions from the user request
2. checks whether the base metric already exists
3. if needed, uses the Metric Creation skill to create the base metric first
4. creates scenario metrics in MAQL using arithmetic changes, such as multipliers or constants
5. reuses the existing MAQL creation rules already used by the platform
6. creates a comparison visualization, usually in a single chart
7. if data sharing is enabled, compares the scenarios in natural language and recommends which scenario to consider

Scenario metrics can be reused in different analytical contexts, such as different attributes, breakdowns, or time views. Depending on the workflow and user permissions, these scenario metrics may be persisted or used only for the current analysis.

Data Sharing

The What-If Analysis skill always respects the AI Assistant data sharing setting.

When data sharing is disabled, the assistant can still create the scenario metrics and display the comparison visualization, but it does not use scenario results for natural language interpretation.

When data sharing is enabled, the assistant can also state the assumptions behind each scenario, compare the results, and provide a recommendation based on the output. This matches the current agentic AI Assistant behavior, where advanced result interpretation depends on the data sharing setting.

Examples

Compare Discount Scenarios
The user asks: What if we discount by 10% vs 20%?
The AI Assistant then:

• identifies the base metric, such as Revenue,
• creates scenario metrics, for example Revenue × 0.9 and Revenue × 0.8,
• shows the comparison in a visualization, and
• if data sharing is enabled, explains the trade-off and recommends which scenario to consider.

Compare Growth Assumptions
The user asks: Compare revenue with 5% vs 10% growth assumption.
The AI Assistant then:

• creates the scenario metrics,
• visualizes both scenarios together, and
• if allowed by data sharing settings, compares the outcomes and summarizes the difference.

Create a Base Metric First
The user asks: What if gross margin improves by 5% vs 10%?
The AI Assistant then:

• checks whether a matching gross margin metric already exists,
• if needed, creates the base metric first using the Metric Creation skill,
• creates the scenario metrics, and
• shows the result in a comparison visualization.

More Example Prompts

• Compare revenue if prices increase by 5% or 10%.
• Show profit under different discount assumptions.
• Compare sales under low-growth and high-growth scenarios.
• What happens if costs decrease by 5% vs 15%?

Limitations

• This skill builds on metric creation and uses the same MAQL creation rules as the platform
• Scenarios must be expressible as metric formulas using arithmetic operations
• The base metric must exist or be created first
• Scenario metrics may require permission to create metrics in the workspace
• If the user does not have permission to create persistent metrics, the assistant explains the limitation
• Natural language comparison and recommendation are available only when data sharing is enabled
• Some business scenarios cannot be expressed as metric formulas alone

Error Handling

If the assistant cannot create or compare the requested scenarios, it explains the issue and suggests a next step.

Examples include:

• the requested scenario cannot be expressed as a metric formula
• the base metric does not exist and cannot be created
• the user does not have permission to create persistent metrics
• the scenario definition is too ambiguous and needs clarification

In these cases, the assistant may suggest simplifying the scenario, defining the base metric first, or rephrasing the request using arithmetic assumptions that can be represented in MAQL.