Direct relevance validation uses a constrained binary prompt across repeated sampling to calculate a Query Relevance Score based on the proportion of affirmative recommendations an AI model returns for a brand against specific queries or entities.