For faculty

Process-based writing assessment: grading the draft, not just the essay

The finished essay is the smallest part of what a student did. Assessing the process answers the integrity question and the feedback question at the same time.

Last reviewed September 9, 2026. Written by the EssayGuard team.

In short

Process-based writing assessment grades how a piece of writing was made alongside what it says. The instructor reads a record of the drafting, such as a session replay, and scores planning, revision and use of sources as part of the mark. It answers the integrity problem, because generated text has no drafting process, and the feedback problem, because the record shows where the writing went wrong.

The practice

Process-based writing assessment grades the drafting alongside the draft. The student still hands in an essay, and the essay is still read and marked. But part of the grade rests on a record of how the essay was built: the planning, the false starts, the paragraphs rebuilt, the sources brought in and reworked. The instructor reads that record the way a studio teacher walks past an easel, and the feedback follows from what the record shows.

Writing programs have taught process for decades through outlines, annotated bibliographies and staged drafts. What changed is that the record can now be the actual writing session rather than a set of artifacts produced around it. A session replay shows the draft being written, keystroke by keystroke, with the pauses and pastes in place. Nothing has to be handed in separately, and nothing has to be reconstructed afterward.

Why it answers two problems at once

The integrity problem and the feedback problem look unrelated, and process assessment happens to answer both.

On integrity: generated text has no drafting process. It arrives in one paste, or it is retyped from another window in a rhythm that does not look like composition. When part of the grade depends on the record showing a piece of writing come together, the shortcut stops producing a full mark on its own, and the class settles into doing the work because that is what is being assessed. This is not detection. No prose is scored, and no verdict is issued. The assignment simply asks for the process, and the process is either there or it is not.

On feedback: a comment on a finished essay can only be about the finished essay. The record shows the decisions behind it, and the decisions are where the learning is. An instructor who can see that a thesis was rewritten four times before the introduction was touched can say something specific about how this student plans, which is advice a finished document never supports.

What a replay shows

A replay of a recorded session shows the essay being written at the speed it was written, or faster if you scrub. Four things stand out within a minute or two.

  • Where the argument stalled. Long pauses cluster at the points a writer did not yet know what to say. If they cluster before the second body paragraph every time, that is where this student runs out of plan.
  • Which paragraph was rebuilt. Heavy revision density in one place shows the passage the student fought with. Sometimes that is the strongest paragraph in the essay and sometimes it is the weakest, and both are worth a comment.
  • Typed versus pasted. Pasted text stays marked for the rest of the session. A quotation pasted from a source and then worked into a sentence looks different from three paragraphs pasted and left alone.
  • The order the thinking formed. Some students write the conclusion first. Some write the body and back into a thesis. The order is a fact about how this writer works, and it is usually invisible to them.

A document history from Google Docs or Word does not show these things. Both record document states rather than keystrokes, and neither distinguishes typed text from pasted text. A keystroke-level record shows the typing rhythm, the pauses, the deletions and the pastes with their length, which is the difference between knowing a document changed and seeing how it was written.

Source: learn.microsoft.com Q&A on Word version history, checked 2026-09-08.

A simple rubric split

Keep the split simple enough to explain in one sentence: most of the grade is the essay, a fixed share is the process. Score the process on what the record shows was done, not on how long it took. An example for a 1,500-word argumentative essay:

ComponentWeightWhat earns full marks
Product: argument and evidence40A defensible thesis, supported and qualified
Product: structure and prose30Paragraphs that each do one job; sentences that are clear
Process: revision15The record shows at least one substantive rewrite, not only proofreading
Process: use of sources10Quoted material is pasted or typed, then integrated rather than left standing
Process: reflection5A short note on what changed between the first sitting and the last

Under that rubric a polished essay that arrived in a single paste with no revision can still score 70. It cannot score above 70, and the student knows that in advance. A rough essay with a visible, honest drafting record can score well on process and get specific advice on product. Both outcomes are the ones you want.

Feedback from a replay in five minutes

Five minutes per student is enough once the routine is set. The order matters.

  1. Open the timeline, not the replay. Note the number of sittings, the total writing time and where the revision density is highest. One minute.
  2. Jump the replay to the densest revision. Watch thirty seconds of it. Write one sentence about what the student was trying to fix and whether they fixed it. Two minutes.
  3. Jump to the longest pause. Look at what was written right before it and right after. Write one sentence about where the plan ran out. One minute.
  4. Check any paste over a sentence long. If it was integrated, say so. If it was left standing, say that. One minute.

Three sentences of feedback that come from the record are worth more than a page of margin comments, because the student can see the moment you are describing. If the student can open their own replay, point them at the timestamp.

Setting it up

Put the rubric in the syllabus and in the assignment prompt, with the weights. State that writing for the course happens in a recorded editor, what is recorded and what is not, and how a student who cannot use the editor asks for another arrangement. The syllabus language page has three blocks you can paste. In the first week, have the class write one ungraded paragraph in the editor and then show them a replay, so the first graded piece is not the first time they see the recorder.

For EssayGuard specifically, create the assignment, set the retention window, and share the join link. Students open it in any browser and write. There is nothing for them to install.

Limits

A process record shows how text entered the editor. It does not show where the ideas came from, whether a friend dictated the outline, or whether a student typed from a printout of generated text in another room. A slow, revised, honest-looking session is consistent with the student writing it and also consistent with a careful transcription. The record narrows the question and gives a conversation something to stand on. It does not close the question, and it should not be graded as if it did.

Process assessment also rewards a particular kind of writer, the one who revises on the page. Students who plan thoroughly before typing leave a thinner record, and the rubric has to leave room for them: score the presence of decisions, not the volume of edits. And keep the process share modest. The essay is still the point.

Questions people ask

Does grading the process penalize students who write fast?

Not if the rubric scores decisions rather than duration. A student who plans in their head and drafts in one sitting still leaves a record of sentences abandoned, paragraphs moved and passages tightened. Score whether the revision happened and what it did, never how long the session ran.

How much of the grade should the process carry?

Enough that skipping it costs something, and not so much that the essay stops mattering. The example on this page gives process 30 percent. Start lower for a first term if you like, and adjust once you have read a few dozen replays.

Do I have to watch every replay?

No. Open the timeline view, which summarizes sittings, revision density and pastes on one screen, and watch the replay only where the timeline shows something worth a closer look. A five-minute pass per student is realistic for a class of thirty.

Can a student see their own process record?

Yes. In EssayGuard the student can view and export their own session. Showing a student their own replay is one of the better feedback conversations available, because the evidence for the advice is on the screen between you.

Is a process record the same as an AI detector?

No. A detector reads finished prose and estimates whether a machine wrote it. A process record shows how the text entered the editor and makes no estimate about the prose at all. EssayGuard records the session and does not score writing for style or likely authorship.

Try it with one assignment

The rest of this semester costs nothing. Setup takes three minutes.

Every course you teach, unlimited students. Students install nothing.