Enterprise AI Measurement Guide
Team Performance
Prompt Quality
How good are our engineers at prompting AI coding tools, and how much productivity are we leaving on the table because of gaps in prompting skill rather than gaps in tool capability?
What it shows
Prompt Quality is a scored measure of how well engineers construct their prompts when working with AI coding tools, evaluating structural quality, specificity, context provision, and the degree to which the prompt constrains the AI toward a useful output. The score runs from 0 to 100 and reflects the engineering team's aggregate prompting skill, which Larridin measures from behavioral telemetry rather than self-report.
Why it matters
Prompt Quality is the most underinvested lever in AI coding ROI. An engineer using a capable AI tool with poor prompting practices will consistently get worse results than an engineer using the same tool with strong prompting discipline, yet most organizations measure AI tool adoption and assume skill is developing on its own. A Prompt Quality score of 62 with Use Case Diversity of 24 is not a tool problem. It is a training problem. Knowing that changes where the ROI investment goes.
The Larridin angle
Larridin measures Prompt Quality from structural and metadata signals on the prompts engineers send, without reading prompt content, rather than from survey responses or manager assessment. That makes the score a reliable basis for training program design rather than a rough directional estimate.