Guide

Referential integrity check: find orphaned records

A referential integrity check confirms every record has its parent key — no orphaned rows where a child points at a parent that doesn't exist. Point it at your datasets and every orphan is flagged, with the real failing rows.

No orphaned rows: validate that every foreign key resolves to an existing parent across your tables or files, and see exactly which records break the link.

How it works

1 · Point at the dataThe child and parent datasets (files or connectors).

2 · Check keysEvery child key is matched against the parent key set.

3 · Flag orphansRecords whose parent is missing are listed with examples.

What you get

Orphan detection — every child record without a valid parent flagged.
Within- or cross-file — check referential integrity inside a file or across tables.
Real failing rows — see the exact orphaned records, not just a count.
AI-suggested relationships — the AI can spot likely parent/child key links.
Part of data quality — integrity is one of the scored DQ dimensions.
Auditable — exportable evidence for controls and sign-off.

Referential integrity without writing SQL joins

Checking referential integrity by hand means writing anti-join SQL for every relationship and eyeballing the output. A referential integrity check resolves every child key against the parent set automatically and lists the orphaned records with examples — repeatable and auditable.

This is one of 20+ checks in AI data validation — the same engine also runs reconciliation, data quality, profiling, document compare and governance. For sensitive data, the full dataset never leaves your server; see the security page.

FAQ

What is a referential integrity check?

It verifies that every record's reference (foreign key) points to an existing parent record — no orphans. AI DataValidator checks this across your datasets and lists the exact orphaned rows, with no code.

Can it check integrity across two different files or tables?

Yes — validate that child keys in one dataset exist in the parent dataset, or check relationships within a single file.

Does it show which records are orphaned?

Yes. You get the real failing rows (examples), not just a pass/fail count, so you can fix them.

Do I need to define the relationships?

You can specify them, or let the AI copilot suggest likely parent/child key links from the data.

Run a referential integrity check free

Find every orphaned record across your datasets — with the exact rows that break the link.

Start free — 7 days, full access