The What Why and How of Feedback
Narrative Feedback (NF) contains contextual and constituting information regarding performance that goes beyond feedback via ratings or rankings. Differences in this important part of feedback can either help or harm employees improve their performance. We rely on archival data from the performance management system of an ecommerce company to investigate the effect of NF on performance. Using natural language processing and textual analysis, we find that, on average, the level of detail in NF concerning strengths and the level of causal language use in NF concerning weaknesses of the employee both increase the likelihood of performance improvements. We find no overall effect of NF sentiment, but a negative effect of positive sentiment in NF concerning strengths on the likelihood of performance improvement for low performers. Our findings contribute to the feedback literature by showing which aspects of NF employees respond to and how, improving our understanding of the factors that make feedback valuable.