write-sql-query · gpt-4o · data · 90% confidence

Write a SQL query with GPT-4o — optimized prompt

A GPT-4o SQL prompt: schema as DDL in a fenced block, the question stated plainly, output is the query plus a one-line rationale and the assumptions made.

90% confidenceUpdated 9/7/2026write-sql-query · gpt-4o · data

Original prompt

Write me a SQL query that gets the top customers by revenue this year.

Optimized prompt

Act as a SQL engineer. Write one query for the target dialect and nothing else around it.

Dialect: [Postgres / MySQL / BigQuery / ...]
Schema:
```sql
[PASTE CREATE TABLE STATEMENTS or column lists with types]
```
Question: [what the result set must contain, and the grain — one row per what?]
Filters: [date range, status, tenant, etc.]

Return:
1. The query, formatted, as a fenced sql block.
2. Assumptions — one line each (which column is the join key, how NULLs are treated, tie-breaking).
3. One line: what index or partition this leans on.

No explanation paragraph, no sample output. If the schema is missing a column you need, name it and stop.

Why this GPT-4o prompt works

Tuned for GPT-4o: schema as DDL in a fenced block, the grain stated plainly, output limited to the query plus one-line assumptions. GPT-4o writes correct SQL fast when the target dialect and join keys are pinned and it is told not to explain.

What to check in the output

The generated SQL query correctly joins 'orders' and 'customers'.
The generated SQL query correctly filters by 'order_date' > '2023-01-01'.
The generated SQL query correctly filters by 'total_amount' > 100.