AI Chart Recommendations

When you ask a question in Conversational Analytics or generate a report, the AI automatically picks a chart type. Here is how it decides and how you can override it.

How the AI Chooses Charts

The AI considers several factors when selecting a visualization:

On dashboards, additional chart types are available including pie charts, scatter plots, heatmaps, and gauge charts.

Overriding the AI Choice

You can explicitly request a chart type in your question:

"Show me a line chart of monthly revenue" (forces line chart)
"Create a pie chart of market share by brand"
"Display this as a horizontal bar chart"
"I want a scatter plot of price vs. units sold"

The AI will respect your request even if it would normally choose a different type. If the requested chart type genuinely cannot work with your data (e.g., a pie chart for time series data), the AI will explain why and suggest an alternative.

When the AI Gets It Wrong

Sometimes the AI picks a suboptimal chart. Common issues:

• Too many pie slices — pie charts with 10+ categories are hard to read. Ask for a bar chart instead.
• Line chart for non-time data — line charts imply continuity. If your categories are not sequential, a bar chart is better.
• Missing the right aggregation — the AI might show raw values when you wanted averages or totals. Be explicit: "average order value" vs. "total orders".

Chart Recommendations in Dashboards

When using the AI-assisted dashboard builder, the AI recommends widgets based on your data source. Recommendations consider:

• Column data types (numeric, categorical, date)
• Number of distinct values per column
• Dashboard capacity constraints (up to 20 widgets per dashboard)

You are always free to ignore recommendations and add whatever widgets you prefer.

Related

• Widget Types Reference
• Writing Better AI Queries
• Conversational Analytics