When the number of rows a table gains (or loses) breaks from its normal pattern — too many, too few, or none.
Reviewed by Francisco Ferreira ·
A row count anomaly is when a table's volume moves in a way it normally doesn't: a daily load that usually adds about 10,000 rows suddenly adds 200, or zero, or a million. Zero often means a broken pipeline; a flood often means a duplicate load; a slump often means a partial failure upstream.
Row count is the cheapest, highest-signal health check there is — you don't need to read the data to know something's wrong when the volume is off. It catches broken loads that pass every other check because the rows that did arrive are individually fine.
Tabkeel learns each table's expected intake (including weekday and weekend rhythm) and alerts when the count lands outside the normal band.