Every gradebook platform asks the same two questions before the first class meets: what are your categories, and what is each one worth? Most of us answer in five minutes, copy last year's percentages, and get back to planning actual lessons. Then week nine arrives. A grade drops seven points overnight because of one quiz, a parent wants to know why a 96% homework average "barely counts," and the five-minute decision starts generating hour-long conversations.
Here is the premise of this guide: category weights are not admin. They are the course's official answer to the question "what counts as evidence of learning?" Set them on purpose and the gradebook defends itself all term. Set them by habit and you will be defending it yourself, one email at a time.
One scope note before anything else: everything below describes typical setups. If your school, department, or district mandates category weights, minimum assessment weight, or rounding rules, that policy wins — check it before you build.
Start from what the grade should certify
Before you touch a percentage, answer one question in writing: if two students finish this course with the same letter, what should that letter guarantee they can both do?
A grade is a message with several readers — the student, their parents, next year's teacher, eventually an admissions office. If the message you intend is "this student can demonstrate the skills under exam conditions," your weights should say so. If it is "this student did the work of the course consistently," different weights say that. Neither is wrong. What causes trouble is a gradebook that quietly says one thing while you believe it says another.
Write the sentence — "the grade in this course primarily reflects ___" — and keep it visible. Every decision below is downstream of it.
Choose the machinery: points or weighted categories
There is one structural fork before weighting proper: a total-points gradebook or a weighted-category one. The short version: if your assessment plan is fully settled and you want students to run the math in their heads, points is transparent and honest; if the term will evolve and you want each category's share of the grade protected no matter how many items land in it, weighted categories are safer. The two systems can hand the same student noticeably different grades, and the full comparison — with the arithmetic laid bare — is in points-based vs weighted gradebooks.
The rest of this guide assumes you chose weighted categories, because that is where the setup decisions multiply.
Keep categories few and fat
Three to five categories is the workable range. Each category should be a distinct kind of evidence — practice, checks for understanding, applied work, supervised demonstration — not a distinct assignment format.
Two rules keep the structure honest:
- The merge rule. If two categories would always move together — "quizzes" and "exit tickets" that measure the same thing on the same weeks — merge them. Extra categories add explanation load without adding information.
- The fatness rule. A category should expect enough entries that one bad day cannot dominate it. A category worth 20% with a single item in it makes that item 20% of the grade — a landmine early in the term, when the running grade swings wildly, and at the end, when there is no room to recover. If something happens once, either give it its own honestly named category ("Final project — 20%") or fold it into a larger one.
Set the percentages with three rules
Rule 1: supervised evidence carries the most weight. Work you watched happen — tests, in-class writing, presentations, labs — is the evidence you can personally vouch for. Work done at home, in the era of help apps and AI tools, measures diligence and access to help as much as it measures mastery. That does not make homework worthless; practice deserves credit. It makes homework a poor majority shareholder.
Rule 2: formative work weighs enough to matter, not enough to inflate. A homework or practice category somewhere around 10–20% (a typical range, not a rule) pays students for the practice that produces learning, without letting compliance alone manufacture a letter grade. If practice can outvote demonstration, expect grades that surprise everyone on exam day.
Rule 3: every number must survive being said out loud. "The final exam is 25% because it is the only cumulative, supervised demonstration in the course" survives. "Homework is 40% because I assigned a lot of it" does not. If you cannot finish the sentence "this category is X% because…" for every category, the weights are not done.
The same scores under two weightings
Here is how much these choices move a real grade. One student, five category averages: Homework 92, Quizzes 78, Projects 85, Midterm 74, Final 80.
Scheme A — demonstration-forward: Homework 15%, Quizzes 20%, Projects 25%, Midterm 15%, Final 25%.
- Homework: 92 × 0.15 = 13.80
- Quizzes: 78 × 0.20 = 15.60
- Projects: 85 × 0.25 = 21.25
- Midterm: 74 × 0.15 = 11.10
- Final: 80 × 0.25 = 20.00
- Course grade = 81.75%
Scheme B — practice-forward: Homework 30%, Quizzes 20%, Projects 20%, Midterm 10%, Final 20%.
- Homework: 92 × 0.30 = 27.60
- Quizzes: 78 × 0.20 = 15.60
- Projects: 85 × 0.20 = 17.00
- Midterm: 74 × 0.10 = 7.40
- Final: 80 × 0.20 = 16.00
- Course grade = 83.60%
Same student, same work: 81.75% versus 83.60%. On many typical letter scales that gap crosses a boundary — check the bands in grading scales explained, and remember your school's published scale is the one that counts. Neither scheme is an error. Scheme A weights what happened under your supervision; Scheme B pays consistency and effort. The setup question is which claim you intend the grade to make — which is why the sentence you wrote in step one matters.
Write down the policies your weights imply
Weights come with consequences that only show up mid-term unless you decide them now:
- Dropped scores. Dropping the lowest quiz raises the effective weight of every remaining quiz. That is usually fine — but say it in the syllabus, because students will do this math in November.
- Extra credit. In a weighted book, bonus points only move their own category. Five bonus homework points move the final grade by 5 × the homework weight — under Scheme A above, 0.75 of a point. Decide whether extra credit exists, where it lives, and what it can and cannot rescue.
- The early-term display. With one quiz entered, the quiz category is that one score, and the running grade can look brutal or inflated. Decide what students see in the opening weeks — many teachers simply warn the class that the number stabilizes as categories fill.
- Late work and retakes. Whatever your policy, apply it inside the category, consistently, and write it down. Grading disputes are almost never about the math; they are about a policy that was applied but never announced.
Pressure-test before you publish
Invent three students and run each through your weights: a steady B student, an exam-strong student who treats homework as optional, and a homework-perfect student who freezes on tests. If the ranking your scheme produces is not the ranking you would defend to a colleague, adjust the weights now — changing them mid-term ranges from awkward to prohibited, depending on your school's policy.
Then read the finished syllabus the way your sharpest student will on day one. How to judge a grading scheme walks through that audit from the student's side of the desk — reading your own course through it is the fastest edit you will ever make.
Hand students the math
The final setup step is the one that saves the most time later: publish the formula and teach it once.
Course grade = (category average₁ × weight₁) + (category average₂ × weight₂) + … for every category.
Point your class at the step-by-step weighted grade walkthrough or the fuller weighted grades explainer — both are written for students. A student who can compute their own standing, and work out what they need on the final without you, sends fewer panicked emails and makes better plans. For parents, one worked example in the syllabus — real numbers, every step shown — defuses more conferences than a page of policy prose.
FAQ
How many categories should a gradebook have? Three to five. Fewer than three usually means unlike evidence is being averaged together; more than five means some categories are too thin to absorb a bad day, and the structure gets harder to explain than the grades themselves.
Do the weights have to add up to 100%? Yes. If your platform lets weights sum to something else, it silently rescales each category by weight ÷ total — your "20%" category may not be worth 20% at all. Keep the arithmetic where everyone can see it: make the weights sum to exactly 100.
Should the final exam carry the most weight? Not automatically. Weight it heavily if a cumulative, supervised demonstration is genuinely the best evidence of what your course teaches — and check whether your school sets a minimum or maximum weight for final assessments. A skills-based course with strong project work can defensibly keep the final modest.
How does extra credit behave in a weighted gradebook? It only moves the category it is entered in, scaled by that category's weight — bonus points in a 15% category move the course grade by a fraction of their face value. If you offer extra credit, tell students exactly where it lands so nobody banks on a rescue the math cannot deliver.
What do I tell parents who ask why homework "doesn't count"? That it counts exactly the amount printed on the syllabus, and show the two-line calculation: homework average × homework weight. Then show the same line for the categories that weigh more. The worked example ends the argument in a way the policy paragraph never does.
Weights chosen on purpose survive contact with the term. Write the one-sentence claim your grade makes, keep categories few and fat, give every percentage a reason you can say out loud, and pressure-test the scheme before students ever see it. When you want to check the arithmetic — or hand students a way to check their own — run any weighting scheme through the free grade calculator on AddThisMarks; it computes instantly in the browser, no sign-up needed.
