Paste-the-quiz & rubrics
Tally has no answer-key spreadsheet. You paste the quiz once, it works out what kind of questions you have, and it drafts the rubric for the ones that need judgment.
What to paste
The questions themselves, numbered. Tally recognizes 1., 1), and Q1:. You don't need to clean up formatting, and you don't need to mark which questions are which type.
2. Name two costs that stay fixed as output rises.
3. Explain why cash flow can be negative in a profitable month.
Question 1 has inline options, so Tally treats it as multiple choice and asks you to mark the correct one. Questions 2 and 3 have none, so they're short answer — and those are the ones that get drafted key points.
Key points, and why you should edit them
For each short answer, Tally drafts three to five specific things a full-credit answer must contain. For question 3 above, that's roughly:
This is the highest-leverage minute you'll spend. The draft doesn't know what you emphasized in class — if you spent a week on receivables and barely mentioned examples, delete the last chip and the grading follows your teaching. Tap any chip to edit it, or add your own.
Save it when you're happy. A rubric you've tuned is worth keeping — save it privately to reuse next term or across sections, and optionally submit it to the shared library.
Reviewing flags & overrides
Tally doesn't ask you to check twenty-six answers. It asks you to check the two or three it wasn't confident about, and shows you why.
What gets flagged
Overriding a score
Expand any row to see the student's answer, the score, and the reasoning behind it. The − and + buttons adjust the score; Accept marks the row ready. You can override any answer, not just flagged ones.
Feedback text is editable too. If you rewrite it, the student sees your words, not the drafted ones.
Nothing publishes on its own. Publishing writes draft grades to your Classroom gradebook. Students see nothing until you return the assignment in Classroom, exactly as with hand grading.
If grading stops partway
Chrome sometimes suspends background work on an idle machine. Tally saves a checkpoint after every batch — reopen the popup and it resumes where it stopped. No answer is graded twice, and nothing already reviewed is lost.
Grading scale explained
Short answers: 0, 3, or 5
Three bands, deliberately. A finer scale sounds more precise but produces less consistent grading — the difference between a 7 and an 8 out of 10 isn't defensible, while the difference between "got it," "partly got it," and "didn't" is.
Override any of these to any value your gradebook accepts — the three bands are Tally's starting point, not a limit on you.
Multiple choice: exact match
Case-insensitive exact match against the option you marked correct. No AI involved, no ambiguity, instant. Points come from the question's own value rather than the 0/3/5 scale.
Essays: weighted criteria
Scored out of a total you set (20 by default), across criteria you weight yourself. See essay grading.
Essay grading
Longer writing needs a different rubric and a stronger model, so essays run on their own track — and are metered separately from short answers.
What's different
- Four feedback strands instead of one paragraph — thesis, evidence, organization, and mechanics are addressed separately.
- Weighted criteria you define, rather than flat key points. Build these in the rubric generator.
- A holistic score out of your chosen total, not a 0/3/5 band.
- Specific citation. Feedback quotes or paraphrases a line from the essay rather than generalizing.
Limits
Essays are capped at 20,000 characters each — roughly 3,000 words, well beyond a typical high-school assignment. The free tier includes 20 essays a month, tracked separately from your 200 short answers; see pricing for the essay add-on.
Worth saying plainly: essay grading is the place to read the feedback before publishing, not just the scores. Longer writing has more ways to be unusual, and your judgment on a borderline essay is worth more than it is on a two-sentence answer.
Originality checks
Any short answer can be checked for likely AI-generated or copied text. The result is information for you — never an automatic penalty.
What you get back
A verdict, a confidence level, a plain-language reason, and the underlying request and response data — so you can see the reasoning rather than trusting a number from a black box.
How to read it honestly
AI-detection is probabilistic and it is wrong sometimes — in both directions. It is more likely to misjudge multilingual students and students who write in a formal register. Treat a flag as a reason to look closely or start a conversation, never as evidence on its own.
Tally will never lower a grade because of an originality result. The check is advisory, it's surfaced with its reasoning, and the decision is entirely yours. We built it this way deliberately.
Standards tagging
Tagging a rubric to a standard makes it findable — by you next term, and by colleagues teaching the same unit.
Supported sets
Common Core, state standards, NGSS, AP / College Board, CTE frameworks, and district-custom codes. Tags are free text, so a code your district invented works exactly as well as a national one.
Conventions worth following
Use the official code as written — CCSS.ELA-LITERACY.W.9-10.1, NGSS.MS-LS1-2 — rather than a description. Codes are what other teachers search for, and consistent formatting is what makes a shared library usable rather than a pile.
Multiple tags are fine and often correct; a good argumentative-writing rubric legitimately addresses several writing standards at once.
What tagging does not do
Tags don't change how anything is graded. They're metadata for finding and sharing rubrics — the grading follows your key points, not the standard code.
Request a DPA
Tally will execute a district Data Processing Agreement, including state-specific addenda. See the DPA terms and request page, or email help@gradewithtally.com with your district name and the agreement form you use.
What to include
- Your district's standard DPA, or a note that you'd like ours.
- Any state addendum required — NY Ed Law 2-d, Illinois SOPPA, Texas Student Data Privacy Agreement, California SOPIPA.
- Whether you're joining a consortium agreement (NDPA / SDPC), and its identifier.
- Your target deployment date, so we can prioritize.
What makes this straightforward
Most DPA friction comes from data inventories and deletion procedures. Tally's answer to both is unusually short: no student data is retained after a grading run, and student identities never reach our servers at all. The data safety page has the retention schedule and subprocessor list your reviewer will ask for.
You don't need a signed DPA to pilot — the free tier works immediately, and no procurement action is required to try it with one teacher.