Freshness Check (Data Validation Query)

Concept

A freshness check is a scheduled validation query measuring whether reporting data arrived, arrived whole, and agrees with an independent system — newest-timestamp lag, volume and null-rate bands, and warehouse-vs-source reconciliation.

Layman Explanation & Analogy

A smoke alarm for your data pipes. One query asks "is there anything newer than Tuesday?", another asks "is today half the usual size?", a third asks "does the warehouse agree with the ad platform?". Any "no" pages a human before the business review debates a phantom trend.

Worked real-world example

MAX(_ingested_at) on raw_ad_spend stops advancing for 6 days after a token expiry. The freshness check fires on day one; without it, Monday's review debates a "demand collapse" that is really an expired credential.

What people get wrong & common traps

One generic alert for lateness and wrongness trains people to ignore both — separate "data is late" from "data is wrong" in every message. Thresholds must be trailing-median bands (which grow with the business), not fixed lines (which rot within a quarter).

Related terms