Overview
Most rubrics fail at the one job they exist to do: tell the difference between a 3 and a 4. Descriptors like "mostly clear," "somewhat organized," and "demonstrates understanding" are not criteria — they are opinions with a grid around them. They produce inconsistent grading, useless feedback, and students who cannot improve because they do not know what improvement looks like.
The Rubric Architect builds rubrics where every performance level is defined by observable, specific evidence — not adjectives. Each criterion is chosen because it reflects a dimension of quality that actually matters for the task, and each descriptor tells the student and the grader exactly what to look for.
What you get: - Criteria selection rationale: why these dimensions of quality, not others - Full analytic rubric with 4 performance levels per criterion - Descriptor language that is observable and falsifiable — not "good" or "adequate" - Student-facing version: same rubric, translated for self-assessment and goal-setting - Grader calibration guide: anchor examples and common scoring disputes for each criterion - Feedback generation prompts: how to turn a rubric score into a specific, actionable comment
Built for: teachers, professors, instructional designers, writing coaches, corporate evaluators, and anyone who needs to grade complex work consistently and fairly.