Hand off the first round, keep the judgment
Grading isn't a single action. It's a series of steps, and only the first of those can be handed off. Here's what that split looks like in practice.
What it does
- Runs the first round and flags where it's unsure
- Leaves the teacher to assign grades and decide
- Works exclusively with pseudonymized tests
- Shares grading guides within the subject team
- Tracks which mistakes keep recurring in the class
What this explicitly is not
Not automatic grading. The system runs the first round and flags where it's unsure; the teacher assigns the grades. Tests near the pass / fail line always go entirely through the teacher's hands.
Anyone who tells you otherwise is selling you a problem for June — the moment a parent disputes a grade and no one can explain how it came about.
What a grading round looks like
This sequence is walked through during the training day using participants' own real tests.
-
Prepare and pseudonymize the tests
Use student numbers instead of names. For digital tests, this means an export from the school platform; for paper tests, a scanning step. This is often the step that takes the most effort to set up properly — once.
-
Write the grading guide
The grading model you already carry in your head gets written down explicitly here: what counts as correct, what counts as partially correct, how points are distributed. This document is the real work, and it's reusable for every batch after this one.
-
Run the first round
The batch runs through in one pass. The result isn't a list of grades but a proposal per answer, flagged wherever the model was unsure.
-
Review and assign grades
You go through the flagged cases and spot-check the rest. Everything near the pass / fail line goes entirely through your hands. You set the grades yourself.
-
Adjust for the next batch
Wherever you had to correct often, the grading guide isn't quite right yet. That adjustment takes ten minutes and makes the next batch noticeably better. This is how it becomes real time savings — after three batches.
-
The error library
Which mistakes keep recurring across a whole class? That's teaching information you'd otherwise only sense, and now actually see. For many participants, this turns out to be more valuable afterward than the hours saved.
Why taking tests digitally makes the difference
The first round can only read what's legible. If a test is on paper, a scanning step gets added, and that's where the weakest link sits — not the technology, the handwriting.
Anyone who grades daily already knows this. Less is written by hand and more is typed, and it shows on the pages. Part of every stack today costs more time to decipher than to assess.
You already have this problem
A student who writes the correct answer illegibly loses points to their handwriting instead of their knowledge. That's true today, without any technology involved. Taking tests digitally solves that — and the fact that this method only works well after that is the second effect, not the first.
That's why the first question in the intake conversation is about digital testing. Not because paper is impossible, but because the answer determines how much you'll get out of this.
Where this doesn't apply
In primary education, handwriting itself is a learning goal. There, a paper test belongs and this conversation doesn't apply. The same goes for work where doing it by hand is the point: sketches, constructions, formulas built up step by step.
Taking tests digitally raises its own questions too — devices, supervision during the test, what happens if the network goes down. Those are decisions for the school, not part of this offering. But they do need to be on the table before anything changes.
In colleges and universities, it's different — and heavier
One course, several hundred students, one deadline for submitting grades. The stack is an order of magnitude larger than in a secondary school, and the time between hasn't grown to match.
At the same time, the core of this method is already established there. The first round is often already handed off in higher education — to teaching assistants. The idea that someone else makes a first assessment while the professor keeps the final judgment doesn't need explaining to anyone there.
The real problem is rarely just the time
When multiple assistants grade the same question, they rarely grade it identically. That difference is the best-known source of grade appeals, and it's hard to defend when the grading model was only ever passed on verbally.
A written-out grading guide solves that, regardless of what technology comes after it. It's also exactly the document this method requires anyway.
The line stays the same. The final judgment rests with the examiner, and work near the pass / fail line goes entirely through their hands. What changes is that the first round happens faster, and that everyone involved starts from the same document.
Practically, there's one difference in your favor: exams are more often already digital. The hurdle that most often blocks progress in compulsory education has usually already been cleared in higher education.
Why training, and not just a manual
The technology itself takes twenty minutes to explain. What takes time is the judgment: which answers you must not hand off, how to write a grading guide that also works for your colleague, and how to recognize when a proposal looks convincing but is wrong.
That last part is the real risk. A wrong proposal doesn't look wrong — it looks just as polished as a correct one. Anyone who doesn't learn to spot that will approve what they should reject. No one learns that from a document.
Where the line is absolute
Tests that determine the difference between passing and failing go entirely through the teacher's hands. No spot-checking, no accepting a proposal as-is. That's not a recommendation, it's a rule, and it's made explicit on the day itself.
What the teacher gains
The hours saved are the first thing people notice. But what participants mention most often afterward is something else: seeing for the first time which mistake the whole class makes, instead of just suspecting it. That information was always in the stack — there just was never time to pull it out.
A short conversation is enough
Tell us where your team loses time today. Then you'll know whether this fits — or whether it's still too early.