GRANULE

GRANULE — data recon in 7 checks
Letter	Check	The question	Excel move
G	Grain	"One row = one what per what?" — before any join	Pivot the business key → Count. Never the surrogate ID.
R	
Repeats & Relations	.
Duplicates in every table you aggregate; orphan keys; is the dimension key unique?	=COUNTIFS() on business key; =COUNTIF(dim!A:A, key)=0
A	
Amounts & Anomalies
	Min/max sane? Does it pass the physical plausibility test?	Sort numerics both ways; helper ratio column → sort desc
N
	Nulls
	Which columns, how many, do they matter, what do I do?	Filter → "(Blanks)", =COUNTBLANK()
U	
Units	
Currency mix, UoM, scale, numbers/dates stored as text	Per-record compare vs master (see below)
L
	Logic	Business rules within rows and reconciliation across tables	Flag columns; header total vs SUM of lines
E	
Extent	Does the data cover the question asked — per entity, not just overall?	Pivot entity → Min/Max of date; scan for early cutoffs