You are treating difficulty score (KD) as if it were standardized. It is not

You are treating difficulty score (KD) as if it were standardized. It is not
KD means a different thing in every tool. We mapped the same 600 queries across three vendors:
— Tool A median KD ▇▇▇ 34
— Tool B median KD ▇▇▇▇▇ 51 (same queries)
— Tool C median KD ▇▇ 27
— Pairwise rank correlation only 0.61
A query that reads 'easy' in one panel reads 'hard' in another because each weights referring domains, anchor text, and page age differently.
The mistake: porting a KD threshold (say 'go after under 30') between tools.
The fix: recalibrate the threshold per tool against 20 terms you already rank for, then trust the relative order, never the absolute value.
Benchmark of the week: cross-tool KD disagrees by 15 points or more on 1 in 3 queries.