The data-join validation gate

The data-join validation gate

Most pSEO pages are built from joined tables. A bad join silently produces wrong pages: mismatched prices, wrong addresses, an entity attributed to the wrong category. Validate the join before it renders.

The pre-build checks:

1. Cardinality — Owner: data lead. Confirm the join is one-to-one or one-to-many as intended. Gate: an unexpected many-to-many means duplicate rows; halt.

2. Orphan rate — count rows that lost their match in the join. Gate: orphans above 5 percent means a key mismatch; investigate before scaling.

3. Fan-out check — did row count balloon after the join? Gate: output rows must equal expected entity count, not a multiple.

4. Spot audit — manually verify 20 random joined rows against the source. Gate: zero mismatches in the sample.

5. Referential integrity — every foreign key resolves. Gate: no dangling references render.

— A silent fan-out join is how you ship the same entity under 40 URLs
— Wrong data at scale is worse than thin data; it erodes trust and earns complaints

Ship gate: don't render until all five join checks pass on the full dataset.