Rubric Template: How to Create a Rubric, With Free Examples
A free Excel rubric template with must-pass criteria and a fair grade conversion, plus the four types of rubric and workplace examples you can copy.
A rubric is a marking guide that lists the criteria you are judging in a piece of work and describes what the work looks like at each level of quality, so different markers reach the same result.
- Write each rubric descriptor as something a marker can find in the work, such as "follows the policy steps in order", not a judgement such as "good understanding".
- Mark a criterion as must-pass when the work can't pass without it, so a strong score elsewhere can't hide a missing essential.
- Adding up rubric points can let a weak answer pass, so decide the result from the levels, not the points total.
- Test a new rubric on a few real pieces of work with a second marker, and rewrite any criterion where your scores are two levels apart.
Set up your branded LMS and deliver training with AI
A rubric is a marking guide that lists what you're judging in a piece of work and describes what each level of quality looks like. Below is a free Excel rubric template with a completed workplace example and samples for three common tasks. You'll also find the four types of rubric, and steps for writing one that two markers would score the same way.
I'm an instructional designer at Coursebox, and I also work in support, where trainers and assessors ask how to set up rubrics for marking. Coursebox makes an AI rubric generator that drafts a rubric from a task description for you to edit.
Here's a completed rubric for a new team leader's written case study. The answer averages exactly at the standard and would score 79% on points. It still comes out "Not yet", because it misses a step the policy requires.
The free template works out that result for you, and a must-pass criterion can't be averaged away.
Download the blank rubric to mark your own tasks. The example file has the team leader's marked case study with a second marker, the same task as four types of rubric, a guide to writing descriptors, and three sample rubrics, one per tab.
What a rubric is
A rubric tells the learner, before they start, what good work looks like. It then tells the marker how to judge it. Rubrics are used in schools and universities, and in workplace training for written work, projects, role plays and practical tasks.
The three parts of a rubric
| Part | What it is | Example from the team leader case study |
|---|---|---|
| Criteria | The things you judge, one per row | Follows the attendance policy steps |
| Levels | The scale, usually three to five steps | 1 Not yet, 2 Developing, 3 Meets the standard, 4 Exceeds |
| Descriptors | What the work contains at each level, for each criterion | Level 3: private conversation, written record and a review date, in the policy's order |
Most of the work in a rubric is in the descriptors. Criteria and levels take minutes; descriptors that two markers read the same way take testing.
Four types of rubric
The template's Rubric types tab shows the same case study marked all four ways.
| Type | How it works | Best for | Watch out for |
|---|---|---|---|
| Analytic | Scores each criterion separately at each level | Feedback on each part, and must-pass criteria | Slower to mark |
| Holistic rubric | One overall level for the whole piece of work | Quick marking of short tasks | Gives little feedback; one weak part can hide |
| Single-point | Describes only the standard, with space for what fell short and what went beyond | Feedback-heavy tasks and drafts | No score to convert |
| Checklist | Yes or no for each item, with no levels | Steps that are either done or not | Can't show quality |
Use an analytic rubric when the result decides something, such as a sign-off or a grade. Use a checklist when each step is either done or not, as in an observation checklist for a practical task.
Must-pass criteria
Some criteria are essential. A team leader who skips the written record in a lateness case has missed the point of the policy, however well they plan the conversation.
Mark those criteria as must-pass. A must-pass criterion below level 3 means "Not yet", whatever the average.
The answer ignored the workplace context the task asked for. Broad criteria each gave it partial credit, and the credit added up to 66%. Our support team's advice was to mark each criterion separately as Satisfactory or Not Yet Satisfactory, and to write the essential requirement into the criterion itself.
Turning levels into a grade
Points totals mislead in both directions. The team leader's answer scores 19 of 24 points, which is 79% and looks like a pass. But it misses a must-pass criterion.
The opposite happens too. Score "meets the standard" on every criterion, divide by the maximum, and you get 75%. In a system with an 80% pass mark, that's a fail for work that met the standard.
The template decides the result from the levels instead. If your system needs a percentage, it converts so that level 3 on average equals your pass mark, level 4 equals 100% and level 1 equals 0%.
When you don't need a rubric
- The task is a set of steps that are done or not: use a checklist.
- There's one right answer: use a quiz.
- You're judging competence over time, not one piece of work: use observations and a record of evidence across several tasks.
How to create a rubric
The steps follow a written case study from a leadership course. A new team leader plans how to handle a team member's repeated lateness, using the company's attendance policy.
Step 1: Start from the task and its standard
Write down what the learner must produce, and the standard it's judged against. Here, that's a written plan that follows the attendance policy.
Your criteria come from the learning objectives. If they're vague, fix them first with SMART learning objectives.
Step 2: Choose three to six criteria
Pick the things that decide whether the work does its job. The team leader rubric has six:
- Applies the answer to the case (must pass)
- Follows the attendance policy steps (must pass)
- Looks for causes before deciding
- Plans the conversation
- Sets a follow-up
- Writes clearly for a manager
Give each a weight that adds to 100. The policy steps carry the most, at 25%.
Step 3: Write level 3 first
Level 3 is the standard, so write it first. Then write what falls short at levels 1 and 2, and what goes beyond it at level 4.
Each descriptor should describe something you can find in the work. If two markers could read it differently, rewrite it.
Read the descriptor, then pick up a piece of work. If you can't point to the words or actions it describes, rewrite it until you can.
| Vague | Something you can find in the work |
|---|---|
| Excellent understanding of the policy | Follows the policy steps in order: private conversation, written record, review date |
| Good communication | Short paragraphs, plain words, a clear next action |
| Shows critical thinking | Plans questions to find the cause before deciding |
| Some errors | Two of the three policy steps, or the steps out of order |
The team leader rubric, marked
Example: A course facilitator marks a new team leader's written case study on handling lateness.
- 1. Must pass: the policy steps score 2, so the cell turns red. The answer plans the conversation but doesn't record it or set a review date.
- 2. Level 3 column: each descriptor names something the marker can point to in the answer.
- 3. Weights: the six criteria add to 100, with the policy steps worth the most.
Step 4: Test it with a second marker
Before using a rubric for decisions, mark three or four real pieces of work with a colleague. Mark separately, then compare.
Australia's training regulator, ASQA, expects assessment outcomes to be "comparable irrespective of which assessor is conducting the assessment" (ASQA assessment practice guide). A second marker is how you check that.
- Each marker scores alone, with a note on what they saw for each criterion.
- Compare criterion by criterion. The template counts exact agreement and agreement within one level.
- Discuss any gap of two levels or more. It's almost always the descriptor, not the marker.
- Rewrite the descriptor, then mark one more piece of work to check the fix.
For an accredited course, this feeds into assessment validation.
Step 5: Mark, with a note for every score below 3
Score each criterion and write what you saw. Any score below 3 needs a note, so the learner knows what to fix. The template counts a missing note as unfinished marking.
The result for the team leader's case study
Example: The same case study, with the second marker's scores compared.
- 1. Average level: exactly 3.00, the pass average. The points total of 79% would also pass.
- 2. Result: Not yet, because the policy steps are below the standard and they're must-pass.
- 3. Markers agree: 4 of 6 criteria exactly and all 6 within one level, so the rubric is working.
Step 6: Give the result and the feedback
The template gives one of five results:
- Not marked: nothing scored yet.
- Incomplete: a criterion is unscored, or a note is missing.
- Not yet: a must-pass criterion is below the standard, or the average is below the pass average.
- Meets the standard.
- Exceeds the standard: the average reaches the exceeds level you set.
The team leader gets "Not yet" with one clear fix: add the written record and the review date. If the resubmission scores 3 on the policy steps, the result becomes "Meets the standard".
You can add a rubric to an assignment in Coursebox, with criteria, a description for each and performance levels with points. The AI Assignment Writer suggests a rubric from your course content for you to edit. Assignments can be marked Satisfactory or Not Yet Satisfactory, and AI marks can be held until an assessor approves them.
Rubric examples for workplace training
In the 180 days to 19 October 2026, at least 101 people asked our AI course builder for a rubric or marking criteria for workplace or adult training. Role plays, projects and practical tasks came up most.
These three samples cover each kind. Each sample is a tab in the example file, with weights and the must-pass criterion.
A sales role play
A sales rep's practice call with a prospect, marked from a recording or a live role play.
| Criterion | 1 Not yet | 2 Developing | 3 Meets the standard | 4 Exceeds |
|---|---|---|---|---|
| Opens and sets the agenda | Launches into the pitch | Introduces themselves only | Introduces themselves, the purpose and the time needed | Also checks the prospect's own goal for the call |
| Asks discovery questions | No questions | Closed questions only | Open questions about the problem, its cost and who decides | Follows up answers to find the real priority |
| Links the product to what they heard | Generic feature list | Features with no link to the answers | Each point tied to something the prospect said | Uses the prospect's own words and numbers |
| Handles an objection | Argues or drops it | Answers without checking the concern | Asks about the concern, answers it and checks it's resolved | Turns the concern into a next step |
| Makes accurate claims (must pass) | States something untrue about the product | Vague or overstated claims | Every claim is accurate and within what the product does | Accurate, and offers proof such as a case or trial |
| Agrees a next step | Ends with no next step | 'I'll send something through' | A dated next step the prospect agrees to | A dated step with the decision-makers involved |
Accurate claims carry no weight in the score; they're a gate. A rep can't pass by being persuasive about something the product doesn't do. To practise the call first, use these customer service role play scenarios.
A cookery practical
A commercial cookery learner plates a dish to the standard recipe during service.
| Criterion | 1 Not yet | 2 Developing | 3 Meets the standard | 4 Exceeds |
|---|---|---|---|---|
| Keeps food safe (must pass) | Unsafe temperature or cross-contamination | Safe but checks missed | Temperatures checked and recorded, clean boards and hands | Also corrects a risk without being told |
| Follows the recipe | Wrong ingredients or method | Recipe followed with changes | Recipe and quantities followed | Recipe followed and adjusted correctly for the batch size |
| Cooks to the standard | Under or overcooked | Uneven | Cooked as the standard describes | Consistent across every plate |
| Plates to the photo standard | Doesn't match | Matches in parts | Matches the photo: portion, layout and clean rim | Matches every time, at service speed |
| Works to time | Holds up the pass | Late on some plates | Each plate within the service time | Early, and helps the next station |
| Leaves the station clean | Leaves a mess | Partly clean | Clean and stocked for the next person | Clean, stocked and the next prep started |
The quality of a dish varies in ways a yes or no can't capture, so levels work better here than a checklist. Food safety is the exception, which is why it's must-pass.
A cybersecurity report
The final report from a cybersecurity course: a security assessment of a sample small-business network.
| Criterion | 1 Not yet | 2 Developing | 3 Meets the standard | 4 Exceeds |
|---|---|---|---|---|
| Sets the scope | No scope | Scope unclear | States what was tested, how and what was out of scope | Also states the limits of the test |
| Backs findings with evidence (must pass) | Findings with no evidence | Evidence for some findings | Every finding shows how it was found, with output or screenshots | Evidence another tester could repeat |
| Rates the risk | No ratings | Ratings without reasons | Likelihood and impact for each finding, with a reason | Ratings tied to the business's own assets |
| Recommends fixes | No recommendations | Generic advice | A specific fix for each finding, in priority order | Fixes with effort and cost, so the owner can plan |
| Writes a summary for the owner | None | Technical summary only | One page a non-technical owner can act on | Also says what to do this week |
| Presents the report clearly | Hard to follow | Clear in parts | Logical order, headings and plain words | Could be handed to a client as it is |
Six criteria are enough to separate a report that's ready for a client from one that isn't.
What goes wrong when people use rubrics
| What you see | Why it happens | What to do |
|---|---|---|
| Markers score the same work two levels apart | The descriptor uses words like "good" or "some" | Rewrite the descriptor as something you can find in the work |
| Everyone scores 3 | Level 3 describes effort, not the standard | Describe what the work contains at level 3, and test it on weak work |
| A weak answer passes on points | Partial credit on broad criteria adds up | Make the criteria the work can't pass without must-pass |
| Learners are surprised by the result | They saw the rubric after the task | Give the rubric with the task, and show a marked example |
| Meeting the standard converts to a fail | The percentage is points divided by the maximum | Convert so level 3 equals the pass mark, or report the result instead |
Learners can also use the rubric on their own work before submitting. A short self-evaluation against the level 3 column catches most gaps early.
When a rubric isn't the right tool
- To decide when to check progress and when to judge, see formative vs summative assessment.
- For an accredited unit, map the criteria to the unit of competency before you write descriptors.
- To mark at volume, compare the best AI grading tools, and keep an assessor approving the marks.
The template is version 1.0 (20 Oct 2026), tested in Excel for Windows. It uses only functions available since Excel 2010. Check the sample criteria against your own standards before you use them.
Writing your next rubric
Start from the standard, and write level 3 first. Keep to three to six criteria, and mark the ones the work can't pass without. Test the rubric with a second marker before it decides anything.
To draft criteria and descriptors from a task description, start with Coursebox's AI rubric generator, then edit the descriptors until you could find each one in the work.
Frequently Asked Questions
A marking guide that lists the criteria for a piece of work and describes what the work looks like at each level of quality. It tells learners what good looks like, and helps different markers reach the same result.
Analytic rubrics score each criterion separately. Holistic rubrics give one overall level. Single-point rubrics describe only the standard, and checklists mark each item yes or no. Analytic rubrics give the most feedback and suit must-pass criteria.
An analytic rubric scores each criterion on its own, so learners see which parts need work. A holistic rubric gives one overall level for the whole piece, which is quicker to mark but gives less feedback.
Three to five. Four levels (not yet, developing, meets the standard, exceeds) work well for workplace training. They avoid a middle option markers default to, and they make "meets the standard" a clear line.
Yes. Rubrics suit written work, role plays, projects and practical tasks in workplace training.
Carolina Martin
Carolina is a highly experienced learning designer with tech start-up experience. She is based in Sydney and previously designed courses alongside Coursebox founder Travis in a previous job. She is passionate about delivering engaging learning experiences and brings the same perspective to customer experience too. She manages our customer support documentation and product testing processes.
