A mid-market distributor we worked with in Chungcheong listed the same packaging supplier under eleven name variants. Until those rows merged, tail-spend looked artificially fragmented and a strategic vendor appeared mid-tier.
We begin with legal entity names from your master file, then apply fuzzy matching on tax IDs where available. Korean and English trade names run through a manual review queue — automation alone misses honourific suffixes and branch codes.
Parent-subsidiary relationships get explicit flags. Spend rolled up to the parent for concentration charts; subsidiary detail remains available for plant-level questions. Without that distinction, a regional buyer looks over-consolidated while headquarters sees false diversification.
We document every merge in a change log your audit team can inspect. Reversing a match takes one line item, not a full rebuild. That transparency matters when category managers challenge a ranking.
Only after the vendor file stabilises do we attach categories. Otherwise you remap twice and lose confidence in the first readout.