I have a file with couple hundred thousand rows of account clusters. We are using this life for de-duplication of accounts pulled from 4 databases. Each cluster on average is 6 rows of accounts each with a unique cluster number. Within those clusters someone has to choose one winner account marked with a W in a dedicated column, an L for loser account, and D for accounts we don't want to merge. I need to determine if there are clusters of accounts that do NOT have a W in the cluster. Anyone have an idea for this?