Build a Better Rubric
This tool guides you through a research-backed rubric design process — the same thinking a CFE workshop would walk you through. You'll reflect on your assignment, articulate what success looks like, and build criteria and descriptors that are clear, fair, and genuinely useful to your students.
Saving Your Work
This tool does not save automatically. Use the Save Progress button in the sidebar at any time to download a file you can reload later. Your rubric is yours — nothing is stored on any server.
If you are coming back to a saved session, click Load Progress to pick up where you left off.
Assignment Context
Tell us about your course and assignment. This context shapes everything downstream — the criteria we suggest, the language we use, and the best-practice guidance we surface along the way.
Select the type that best describes this assignment. Assignment type shapes which criteria the tool suggests — a lab report and a policy brief call for very different rubric dimensions.
This helps us suggest relevant criteria even for custom assignment types.
Your Assignment Prompt
Paste the directions you give students. A rubric is only as good as the assignment it assesses — so before we build criteria, we want to make sure your prompt is giving students what they need to succeed.
Include everything students see: the task description, any requirements (length, format, sources), and the purpose or audience if stated. The more complete this is, the better the tool can reflect your own language back to you in the rubric.
A rubric can only be as clear as the assignment it assesses. If students don't fully understand what they're being asked to do, even a well-designed rubric won't close that gap. Consider revisiting your assignment directions before continuing — or reach out to the CFE for a consultation.
Does your assignment include a clear statement about how students may or may not use AI tools? This is now considered a best practice for all assignments, regardless of discipline or level.
Without a clear AI policy, students make their own assumptions — and those assumptions vary widely. A statement in the assignment prompt (not just the syllabus) ensures every student sees the expectation before they begin work. If you haven't defined a role for AI in this assignment yet, the Assessment Design Tool can help you think through the Four Roles framework.
→ Open the Assessment Design ToolWhat Does Success Look Like?
Before naming any criteria, let's start with your mental image of excellent work. This is the most important thinking you'll do in this process — and it often surfaces criteria you wouldn't have named if you started with a blank rubric grid.
Think of the strongest submission you have ever received for this kind of assignment — or imagine one. Don't think about what you asked students to do. Think about what the work actually did that made it excellent.
Write freely — don't filter. Specific observations are more useful than general ones. ("The argument was constructed so that each paragraph built on the last" is more useful than "it was well-organized.")
This is often where the most important criteria live — in the gap between good work and excellent work. Many faculty find this harder to articulate than they expect, which is exactly why it's worth doing before you write descriptors.
Try to be behavioral and specific: what does the A student do that the B student doesn't?
What Goes Wrong?
The criteria faculty care about most are often the ones they never made explicit — until they're frustrated reading the same mistake for the fifteenth time. Let's surface those now.
Think about the last time you graded this assignment (or one like it). What were the most common problems? What made you write the same comment over and over?
List as many as come to mind — you can separate them with commas or line breaks.
Sometimes the most important criteria are the implicit ones — the things you assumed students knew without being told. If something consistently frustrates you when grading, it probably belongs in your rubric.
Your Criteria
Based on what you described, here are suggested criteria for your rubric. Each criterion should represent a distinct, assessable dimension of the assignment. Edit, remove, or add your own.
These suggestions are based on your assignment type and what you described in the previous steps. Accept the ones that fit, edit the names to match your language, and add any that are missing.
Scoring Approach
How will you turn your rubric into a grade? Choose the approach that fits your course. You can always use the rubric descriptively for feedback without attaching numbers — but if you need a gradebook score, this is where you set it up.
Performance Levels
Choose how many performance levels your rubric will have and what you'll call them. This decision matters more than it might seem — and research has a clear recommendation.
Rubrics with 4 or 5 performance levels consistently outperform 3-level rubrics in research. Here's why it matters:
- Better inter-rater reliability — raters can meaningfully distinguish between 4 levels. With only 3, the middle category becomes a catch-all that different graders interpret very differently.
- More actionable feedback — students receiving a "Developing" rather than just "Needs Improvement" know there's a middle path, which research shows increases motivation to revise (Andrade & Brookhart, 2026).
- Fewer grade disputes — specificity reduces ambiguity, and ambiguity is what drives disputes. When students can see exactly where they landed and why, conversations shift from "why did I get this grade" to "how do I improve."
- Faster grading — matching work to a descriptor takes less time than writing individualized comments from scratch. More levels means more precise matching, not more work.
- Supports self-assessment — students using rubrics to self-assess perform significantly better when levels are distinct enough to be self-applicable (Andrade, 2019).
Note: Brookhart (2018) recommends against odd-numbered 3-level scales specifically because the middle category is systematically over-selected, reducing discrimination between student performances.
→ Andrade: Using Rubrics to Promote Thinking and Learning (ASCD)Edit the level names to match your course language if needed.
Write Your Descriptors
Now we fill in the rubric grid. Start with the top and bottom levels for each criterion — these are the anchors. The middle levels will be scaffolded from your anchors.
The most common rubric problem is vague descriptor language. Words like "excellent," "good," or "adequate" tell students almost nothing. Strong descriptors are behavioral — they describe what the student does, not how good it is.
- Instead of: "Uses sources effectively" → Write: "Cites at least 3 peer-reviewed sources and integrates each one into the argument rather than dropping quotes without analysis"
- Instead of: "Shows understanding of the topic" → Write: "Accurately applies course concepts to the case and explains why each concept is relevant"
Tip: Start with the top level — your vision of excellent work from Step 03. Then write the bottom level. The middle levels almost write themselves once the anchors are clear.
→ AAC&U VALUE Rubrics — models of strong descriptors across disciplinesPreview & Export
Your rubric is ready. Review it below, copy it to your clipboard, or print it to PDF. Remember — the most effective practice is to share the rubric with students before they begin the assignment.
Go Deeper — Rubric Design Resources
Your rubric is ready.
Share it with students before they begin the assignment — research consistently shows that rubric transparency improves both performance and student confidence. If you want to refine it further or discuss how to use it formatively in class, the CFE is here to help.