This study investigated 299 chat reference interactions from an academic library consortium for instances of teaching and compared these against other characteristics of the chat, such as question content, staff type, user status, user satisfaction, institutional affiliation, length, and shift busyness. Statistical analysis revealed that teaching was more likely to occur when the chat was a research-related question or when the operator was a graduate student worker employed by the consortium but less likely when the operator was a paraprofessional. Chats with teaching were longer but were also associated with higher user satisfaction scores. Teaching rates did not differ by institutional affiliation, shift busyness, or user type. These results indicate that busy consortial services can offer comparable teaching service to single-institution services. The researchers recommend updating operator training to better emphasize teaching to increase user satisfaction.