Docs
Guides
A well-written prompt produces reliable insights, while vague prompts often produce ambiguous or misleading results. This guide explains how to design prompts that clearly communicate:
- Timeframe
- Metric definition
- Business intent
- Required totals
- Expected grouping
Why Prompts Fail
Most AI mistakes occur because prompts are unclear or incomplete. Common issues include:- Missing timeframe
- Vague questions
- Multiple KPIs in one prompt
- Undefined scope
- Missing numerator or denominator
Example of a Vague Prompt
How is performance looking recently?
Typical AI Result
- Timeframe unclear
- No defined metric
- No measurable output
What Makes a Good Prompt
A strong prompt includes:| Component | Purpose |
|---|---|
| Timeframe | Defines the analysis period |
| Metric | Specifies exactly what should be calculated |
| Totals | Requests supporting numbers |
| Context | Explains business intent |
| Segmentation | Defines grouping if needed |
Recommended Prompt Structure
For a defined timeframe, calculate a specific metric and provide the supporting totals required to validate the result.
Prompt Design Examples
Below are several examples showing how better prompt structure improves AI responses.Example 1 — Financial Impact Analysis
Business Intent
Understand the financial impact of a specific operational issue.Poor Prompt
Show the number of failed transactions this month.
Typical Result
- May return only counts
- Financial impact missing
- Date logic may be incorrect
Improved Prompt
For a specific month, calculate the total financial value associated with failed transactions. Include both the number of failed transactions and the total value affected.
Example 2 — Month-to-Month Comparison
Business Intent
Compare operational performance between two periods.Poor Prompt
How did this month perform compared to last month?
Typical Result
- Months may be misinterpreted
- Missing metrics
- No explanation of change
Improved Prompt
Provide the total revenue and total payments for two specified months. Compare the values and explain the percentage difference.
Example 3 — Quality Rate Calculation
Business Intent
Measure the quality of first-attempt submissions.Poor Prompt
What percentage of submissions failed?
Typical Result
- Denominator unclear
- May mix different failure types
- Missing timeframe
Improved Prompt
Out of all submissions within the defined timeframe, calculate the percentage that failed on the first attempt. Provide total submissions and total failed submissions.
Example 4 — Resolution Effectiveness
Business Intent
Measure how effectively issues are resolved.Poor Prompt
What is the resolution rate over the past few months?
Typical Result
- Resolution definition unclear
- Timing logic missing
Improved Prompt
For issues recorded during the last six months, calculate the percentage that were resolved successfully. Use the issue creation date as the reference point and include total issues and total resolved cases.
Example 5 — Activity to Revenue Conversion
Business Intent
Understand how operational activity translates into revenue.Poor Prompt
How well did activity convert into revenue?
Typical Result
- Conversion definition unclear
- Missing timeframe
- No financial breakdown
Improved Prompt
For activities completed within a defined month, provide the total number of activities, total revenue generated, and total payments received.
Example 6 — Risk Exposure Analysis
Business Intent
Identify risk drivers within older balances.Poor Prompt
Why is the outstanding balance high?
Typical Result
- Subjective explanation
- No aging definition
- No breakdown
Improved Prompt
Provide the total balance exceeding a defined aging threshold and break it down by category and responsible entity.
Example 7 — Source Contribution Analysis
Business Intent
Understand the contribution of different revenue sources.Poor Prompt
How much revenue came from customers?
Typical Result
- No timeframe
- No reference total
- Missing percentages
Improved Prompt
For the selected period, calculate the amount collected from customer payments and the percentage of total revenue this represents.
Best Practices for Writing Prompts
Always Include
- Specific timeframe
- Clear metric definition
- Supporting totals
- Business context
- Segmentation if needed
Avoid
- Vague questions
- Multiple KPIs in one prompt
- Undefined metrics
- Subjective wording
Recommended Prompt Template
For a defined timeframe, calculate a specific metric. Provide supporting totals and break down the result by relevant dimensions if required.
Key Takeaway
Good prompts lead to clear, reliable AI outputs. Every prompt should clearly define:- When the analysis occurs
- What metric should be calculated
- How it should be measured
- What totals support the result