Getting the Best Results with Natural Language Data Questions

Be explicit about:

• The metric (e.g. revenue, customer count, profit margin)
• The dimension (e.g. product, region, customer type)
• The timeframe

Examples:

• Good: “What was total revenue for Product A in Q2 2024?”
• Good: “How many new customers joined in EMEA during January 2025?”
• Not good: “How are sales doing?” (too vague)

When comparing values:

• Good: “Did Product A’s revenue grow faster than Product B’s in Q4 2024?”
• Good: “What was the profit margin by region in Q3 2024?”

If a question combines multiple analyses (like filtering, ranking, and comparing), ask it in stages.

Instead of:

Try:

1. “What was the revenue growth by region in 2024?”
2. “Which region had the highest growth?”
3. “Now compare that to the top-performing product’s growth.”

This stepwise approach makes Minds’ answers clearer and helps you follow its reasoning and correct it if needed.

Minds performs best when it understands your company’s terminology. This can be configured in the System Prompt settings.

You can:

• Define business-specific terms that may differ in your business like fiscal year or commonly-used acronyms.
• Reference internal metrics (“Use the ‘net revenue retention’ field we track”).
• Ask it to use the same naming conventions your team does.

This helps Minds interpret your questions exactly as your analysts would.

Minds can respond with more or less context, explanation, and creative analysis depending on your personality and communication preferences. You can shape how it interacts with you, from purely factual to exploratory and strategic.

Define Your Personal Preferences

Tell Minds how you want it to communicate in the System Prompt settings:

• “Be concise and to the point.”
• “Include short explanations with each result.”
• “Use bullet points for clarity.”
• “End each answer with key takeaways.”

This helps Minds match your workflow, whether you need quick answers or detailed insights.

Define the Personality You Prefer

You can set the tone or analytical style that fits your role:

• “Act like a finance analyst focused on precision.”
• “Think like a strategy partner. Include insights and implications.”
• “Respond like a product teammate brainstorming next steps.”

This ensures Minds’ responses sound natural and aligned with your decision-making style.

Define the Level of Confidence and Exploration

Adjust how boldly Minds interprets data:

• “Be conservative. Highlight only statistically clear patterns.”
• “Be exploratory. Propose possible explanations and next questions.”
• “If uncertain, show your reasoning or assumptions.”

This helps balance analytical rigor with creative reasoning.

Examples

Minds excels when questions:

• Filter by date, product, or region
• Ask for top N summaries (count, sum, average, growth rate)
• Compare data across time or categories

You may hit limits if:

• You want to list all occurrences of things that are high in number (e.g. list all of our customers). Minds can provide the SQL needed to retrieve any data directly if needed.
• You include multiple conditions or time periods in one sentence
• The question depends on derived metrics that weren’t defined
• Key filters (like product or region) are missing

When unsure, simplify or ask for the base data first.

You can ask conversationally:

• “Now show me Q3 only.”
• “Break that down by customer segment.”
• “Which region contributed most to that growth?”

No need to restate your original question, Minds keep context as you refine your analysis.

With clear, contextual, stepwise questions, Minds can deliver precise, actionable insights just like a great data analyst on your team.

Examples that work great:

• “What was revenue in North America in Q2 2024?”
• “How many active customers did we have in January 2025?”
• “What was Product X’s profit margin in March 2024?”

Tip: Prefer phrasing like this, it yields highly accurate results.

Examples that benefit from decomposition:

• “Compare customer churn between Q1 and Q3 2024.”

  • Step 1: “What was the churn rate each month in Q1 2024?”
  • Step 2: “Now show Q3.”
  • Step 3: “Compare the two.”

• “What was the average revenue per customer by region in 2024?”

  • Step 1: “What was total revenue and total customers by region in 2024?”
  • Step 2: “Compute average revenue per customer.”

Tip: Ask for underlying data first, then compute or compare.

Examples that can be rephrased:

• “Which region had the highest month-over-month revenue growth in 2024, and what was the value?”

  • Step 1: “What was monthly revenue by region in 2024?”
  • Step 2: “Which region had the highest month-over-month growth?”

• “Which product had the highest profit margin in 2024, and how did that compare to 2023?”

  • Step 1: “What was the profit margin by product in 2024?”
  • Step 2: “Now compare those to 2023.”

Tip: Separate calculations and comparisons into different steps.

Instead of:

Ask:

Be clear about thresholds, filters, and metrics.

Instead of:

Ask: