How to grade your own essay with a rubric before you submit
8 min read · updated 2026-10-14
To grade your own essay with a rubric, read each criterion as a question, find the exact sentences in your draft that answer it, match those sentences to the level descriptors, and write down the score you could defend to your instructor. Then fix the lowest-scoring criterion first and check that row again. The pass catches the point losses that come from missing requirements, which are the easiest points to recover.
Why the rubric beats your gut
You know how much work went into the essay. The rubric does not care. An instructor grading a stack of papers reads each one looking for evidence of specific criteria, usually in the order the rubric lists them. If the evidence is not on the page, the points are gone, no matter how clear the idea was in your head.
Self-grading works because it forces the same shift. You stop asking "is this good?" and start asking "where, exactly, is the thesis?" That question has a yes-or-no answer, and you can act on a no.
Two things you need before you start: the rubric your instructor actually attached to the assignment (not a generic one from a writing center), and a draft that is complete enough to have an introduction, body paragraphs, and a conclusion. Grading a half-finished draft tells you only that it is half finished.
Step 1: Turn each criterion into a question
Rubric rows are written for graders. Rewrite each one as a question you can answer by pointing at your text. Keep the question narrow enough that the answer is a quote, a count, or a no.
Some common conversions:
- "Clear, arguable thesis" becomes "Can a reader write my thesis in one sentence after the introduction, and could someone reasonably disagree with it?"
- "Uses evidence effectively" becomes "Does each body paragraph contain at least one cited piece of evidence, followed by a sentence that explains how it supports the paragraph's claim?"
- "Synthesizes sources" becomes "Is there at least one paragraph where two sources are discussed together to make a point neither makes alone?"
- "Addresses counterarguments" becomes "Is there a paragraph that states an opposing view fairly and then answers it?"
- "Organization" becomes "Does every paragraph open with a sentence that says what the paragraph will argue, and do the paragraphs follow an order I could explain?"
- "APA formatting" becomes "Does every in-text citation have a matching reference entry, and does every quotation have a page or paragraph number?"
Write the questions down. You will reuse them for every paper in the course.
Step 2: Find the evidence in your draft
Do not rate from memory. Open the draft and, for each question, highlight the sentences that answer it. Use a different color per criterion if that helps, or paste the sentences under each question in a separate document.
If you cannot highlight anything for a criterion, the criterion is not met. It does not matter that you meant to include a counterargument or that the thesis is "implied." Graders do not award points for intent.
Count when the rubric implies a number. "Multiple scholarly sources" means you should be able to count them in the reference list and find each one cited in the body. "Evidence in every paragraph" means every paragraph, so count the paragraphs with none.
This step is where the real problems tend to show up. A frequent one is a body paragraph that is pure summary with no claim of its own, or a quotation that sits at the end of a paragraph with nothing after it.
Step 3: Match the evidence to a level, honestly
Now read the level descriptors for each row. Most rubrics have four levels with names like exemplary, proficient, developing, and beginning, or a point scale that maps to the same thing.
Graders read the top-level descriptor as a checklist. If it says "thesis is specific, arguable, and previews the structure of the argument," all three parts have to be present for the top score. Two out of three is the next level down. When you are unsure between two levels, choose the lower one. Your instructor will.
Here is a four-level row for a criterion most rubrics include, with what each level looks like in a real body paragraph. The paragraphs are from an imagined essay arguing that colleges should move final exams off the last week of the term. The authors cited in them are invented for the example.
Criterion: Use of evidence (25 points)
| Level | Descriptor |
|---|---|
| Exemplary (23–25) | Every claim is supported by specific, cited evidence. Each piece of evidence is explained and tied back to the thesis. |
| Proficient (18–22) | Most claims have cited evidence. Explanation is present but sometimes general. |
| Developing (13–17) | Evidence is present but sparse, uncited, or dropped in without explanation. |
| Beginning (0–12) | Claims are asserted without evidence, or evidence is unrelated to the claims. |
Beginning paragraph. "Finals week is stressful for everyone. Students barely sleep and still have to perform at their best. It is obvious that spreading exams out would help. Most people agree with this."
Nothing is cited, and "most people agree" is not evidence.
Developing paragraph. "Finals week is stressful. One study found that students slept less during exam periods (Okafor, 2021). Spreading exams out would help."
There is a source, but the sentence from it is dropped in and left there. The paragraph never says why less sleep matters for the argument.
Proficient paragraph. "Compressed exam schedules reduce sleep. Okafor (2021) tracked sleep in undergraduates across a term and found the sharpest drop in the final week. Less sleep makes it harder to recall material, so a schedule that packs exams together is working against the thing it is supposed to measure."
Cited, explained, and connected to the thesis. The explanation is one general sentence, though, and the paragraph relies on a single source.
Exemplary paragraph. "Compressed exam schedules reduce sleep, and the reduction is large enough to affect scores. Okafor (2021) tracked undergraduates across a term and found the sharpest sleep drop in the final week, when most exams fell. Brennan and Liu (2019) found that students who sat two exams in one day scored lower on the second than on comparable exams taken on separate days. Together, these findings suggest that the schedule itself, not only the material, is lowering performance. That is the gap a spread-out schedule would close."
Two sources, each explained, discussed together, and tied to a specific claim. This is what the top descriptor is asking for.
Step 4: Write one line of justification per row
For each criterion, write the score you chose and one sentence saying why, using the words from the descriptor. For example: "Use of evidence: Developing. Paragraphs 3 and 5 have a source each, but paragraphs 2 and 4 have none, and no paragraph explains how the evidence supports the claim."
Writing the sentence does two things. It stops you from rounding up, because you have to name what is missing. And it becomes your revision list. The sentence above tells you exactly what to do: add evidence to paragraphs 2 and 4, then add an explanation sentence after each quotation.
Step 5: Fix the lowest row first, then recheck
Points are not evenly distributed. Moving a row from developing to proficient is usually worth more than polishing a row that is already at the top. Sort your rows by score, start with the lowest, and do the specific work your justification sentence names. Then reread the paragraphs you changed and score that row again.
Stop when every row is at least at the proficient level or when you run out of time. Do not spend the last hour rewriting a conclusion that already meets the descriptor.
OnRubric does the same pass: paste the essay and the rubric, and it returns a score range per criterion with the sentences it used. Compare its quotes with your highlights. Where you disagree, reread the paragraph. The score is a prediction, not your grade. Your instructor decides, and the tool does not know your instructor. Use it the way you would use a classmate who read the rubric closely: as a pointer to the paragraphs that need work.
Mistakes that make self-grading useless
- Grading from memory instead of from the page. If you did not highlight it, it is not there.
- Treating "meets" as "exceeds." Read the full top-level descriptor before you award the top score.
- Skipping the formatting row because it feels minor. Points lost there are among the fastest to win back.
- Ignoring the weights. A 30-point row deserves three times the attention of a 10-point row.
- Reading the rubric for the first time after the draft is done. Read it before you outline, and turn the criteria into questions then.
- Changing the rubric in your head. If it says "scholarly sources," a news article does not count, even if it is a good article.
Self-grading checklist
Work through this before you submit.