Skip to content
Data & SpreadsheetsIntermediate

Write a SQL Query From a Business Question

Translate a business question into a correct, readable SQL query for your actual schema, with assumptions and checks spelled out.

Share on X (opens in a new tab)Share on LinkedIn (opens in a new tab)

Make it yours

Fill in each variable and watch the prompt update. Anything still in {{braces}} needs your input before you use it.

Fill in the blanks

0 of 3 filled

Your database engine — syntax differs.

Relevant tables and columns, with types if possible.

What you want to know, in plain language.

Live previewUnfilled
You are a careful data analyst who writes clear, correct SQL.

Database: {{sql_dialect}}
Schema (tables and columns):
{{schema}}

Business question: {{question}}

Requirements:
- Use only the tables and columns listed above. If the question can't be answered with them, say what's missing instead of inventing columns.
- Write readable SQL: CTEs for multi-step logic, meaningful aliases, and a comment above each step.
- State every assumption (e.g. how you define "active", how you handle NULLs, time zones, or duplicates).
- After the query, explain in 3–5 bullets what it does, then suggest 1–2 sanity-check queries I can run to verify the result.
Open in Playground

Example

Example input

Filled prompt
You are a careful data analyst who writes clear, correct SQL.

Database: PostgreSQL
Schema (tables and columns):
customers(id, name, signup_date, country)
orders(id, customer_id, order_date, total_amount, status)

Business question: Which 10 countries had the highest average order value from completed orders in the last 90 days, counting only countries with at least 20 such orders?

Requirements:
- Use only the tables and columns listed above. If the question can't be answered with them, say what's missing instead of inventing columns.
- Write readable SQL: CTEs for multi-step logic, meaningful aliases, and a comment above each step.
- State every assumption (e.g. how you define "active", how you handle NULLs, time zones, or duplicates).
- After the query, explain in 3–5 bullets what it does, then suggest 1–2 sanity-check queries I can run to verify the result.

Example output

Illustrative example response — written by Promptly, not generated live
-- Step 1: completed orders from the last 90 days, with the customer's country
WITH recent_orders AS (
  SELECT o.id, o.total_amount, c.country
  FROM orders o
  JOIN customers c ON c.id = o.customer_id
  WHERE o.status = 'completed'
    AND o.order_date >= CURRENT_DATE - INTERVAL '90 days'
)
-- Step 2: average per country, keeping countries with 20+ orders
SELECT country,
       COUNT(*) AS order_count,
       ROUND(AVG(total_amount), 2) AS avg_order_value
FROM recent_orders
GROUP BY country
HAVING COUNT(*) >= 20
ORDER BY avg_order_value DESC
LIMIT 10;

What it does

  • Joins orders to customers to get each order's country
  • Keeps only completed orders from the last 90 days
  • Averages order value per country and drops countries with fewer than 20 orders
  • Returns the top 10 by average

Assumptions

  • Completed orders have status exactly 'completed' (check: SELECT DISTINCT status FROM orders;)
  • Orders with a NULL country form their own group; filter them out if unwanted

Sanity checks

  • Remove the LIMIT and HAVING to see all countries and their counts
  • Spot-check one country: SELECT AVG(total_amount) for its completed orders in the same window

Why it works

Pasting the real schema is the single biggest improvement for SQL prompts — it stops the model from inventing plausible-sounding columns. Requiring stated assumptions exposes the hidden decisions (what counts as "completed", how NULLs are treated) that make queries silently wrong, and the sanity-check queries give you a way to verify the result rather than trusting it.

Techniques used