Essentially, we had to listen to hundreds of records every day, and for each record, we had to answer four questions: 1. Was the person understandable? 2. Was the person actually talking to Siri? 3. Did Siri understand the query correctly? 4. Did Siri respond correctly to the query? See also…