Step 1 — describe what you want to know

Open SQLFix and write your question in plain English. Be specific about the table, the filter, and the aggregation. For example: "total revenue per user who placed more than 3 paid orders in 2025." Specific wording tends to produce a more accurate query than a vague one.

Step 2 — paste a schema if you have one

The schema field is optional, but pasting your table and column names (e.g. users(id, name, created_at); orders(id, user_id, total, status)) helps SQLFix match real columns instead of guessing. If you do not have the schema handy, you can still submit the question and review the result.

Step 3 — pick your dialect

Select PostgreSQL, MySQL, SQLite, or BigQuery. Dialect matters: date functions, string functions, and quoting rules differ between engines. SQLFix writes the query in the dialect you chose, so the output is closer to something you can run as-is.

Step 4 — read the explanation, not just the query

SQLFix returns the SQL in a fenced block followed by a 2–3 sentence plain-English explanation. Read the explanation first. If the explanation does not match what you intended, rephrase your question — small wording changes can change the join or the filter significantly.

Step 5 — check the safety findings

Before copying, review the deterministic safety screen. It flags injection-prone construction, DROP/TRUNCATE without a guard, and UPDATE/DELETE that would touch every row. SQLFix can help you catch these patterns; it does not guarantee the query is correct, so still confirm the logic against your data.

An honest limit

This workflow can help you produce a readable first draft quickly. It does not replace understanding your own schema, and for any statement that writes or deletes data, a qualified reviewer should confirm it before it runs.

Authoritative references

  • SQLite documentation: https://www.sqlite.org/docs.html
  • BigQuery documentation: https://cloud.google.com/bigquery/docs