Process-based writing assessment
Grading only the finished text was always a compromise; it is just that until recently the compromise was cheap. If the product can be produced in nine seconds by anyone, the part worth assessing is the drafting — which is also the part that actually makes writers better. Here is a rubric you can use tomorrow, and the two conversations that come with it.
The idea in one paragraph
Assess what the student did, not only what they handed in. That means a small number of criteria pointed at visible artefacts of the drafting: a plan, a revision made where you can see it, a reflection on what changed, and a record showing the writing was produced over the session rather than arriving at the end. This is not a new pedagogy — process writing has been in English departments for forty years. What has changed is that the process is now the part that carries the evidence, so it has to carry marks too, or students will correctly conclude it is optional.
A four-criterion rubric
Keep it to four. A process rubric that runs to twelve rows is a rubric nobody fills in twice.
- Engagement (0-3). 3: writing is spread across the session, with planning at the start and revision at the end. 2: mostly sustained, with one long gap. 1: a single concentrated burst, little else. 0: nothing produced in the session.
- Revision (0-3). 3: substantial cutting and reordering — text was removed and rewritten, not just appended. 2: some deletion and reworking, mostly local. 1: additive only; the first draft is the last draft. 0: no evidence of revision.
- Reflection (0-3). 3: names a specific change and the reason for it ("I moved the counter-argument up because my second paragraph was answering it before I had said it"). 2: a real change, vaguely explained. 1: generic ("I fixed my grammar"). 0: absent.
- Independence (0-3). 3: the writing was produced in the session; nothing arrived in unexplained blocks. 2: one block worth a question, explained satisfactorily. 1: significant portions appeared at once and could not be accounted for. 0: the session shows no drafting behind the submitted text.
Twelve points. Weight the whole thing at somewhere between a quarter and a third of the assignment, with the finished piece carrying the rest. Below a quarter and students ignore it; above a third and you are grading effort instead of writing.
Two rules keep this defensible. First, every descriptor is written as an observation, never as an inference about intent — "text appeared in one 900-character block" and not "the student pasted". Second, Independence is the only criterion that can ever trigger a conversation, and a low score there is a reason to talk, never a mark you enter and move on from.
What counts as drafting evidence
- A plan or outline, even a photographed scribble.
- Deleted characters. The best cheap proxy for revision: writing that was cut existed once, and someone decided to cut it.
- The distribution of writing over time — whether the words appeared across the session or in one lump at the end.
- The size of insertions. Typing arrives a few characters at a time; a paragraph that arrives at once came from somewhere else, and needs an explanation before it needs an accusation.
- A reflection written in class, at the end, unprepared. Two minutes, and it is the single most informative artefact per minute spent.
- Comparison with the student's own baseline from previous timed pieces. This is more use than any external standard.
Showing it to parents
The parent meeting goes well or badly depending on one choice: show the artefact, or describe it. Play thirty seconds of the drafting and the conversation changes tone, because everyone in the room is now looking at the same thing rather than at your judgment. Three sentences that help:
- "We give marks for the drafting because that is where writing actually improves." The learning reason first, always. Integrity is a consequence, not the pitch.
- "Here is what was recorded, and here is what was not." Name it exactly: the writing in the assignment page, the timing of edits, blocked paste attempts, whether the window lost focus. No camera, no screen recording, nothing about other tabs, nothing outside the editor.
- "No software here decides whether this was written by AI." Say it plainly, because it is the fear underneath most of these meetings. The record shows what happened; the teacher decides what it means.
If a parent asks what happens to the work, have an answer ready with numbers in it: what is stored, for how long, who can see it, and how to delete it.
Showing it to an administrator
An assistant principal wants three things: that the policy is consistent, that it is defensible if challenged, and that it does not create a data problem. So bring one page: the rubric, one anonymised example of a strong session and one that raised a question, the sentence you say to students before every recorded write, and the storage answer. Make explicit that a flag opens a conversation and never closes one — no student is sanctioned on a pattern alone, and there is a documented step between "this looks unusual" and any consequence. Departments that write that step down have a much easier year than departments that do not.
Where WrittenLive fits
WrittenLive records the writing process and replays it. Type an assignment title, get a student link and a private dashboard link, no accounts on either side. Students write in a paste-proof editor; every insert, delete, cursor jump, focus change and idle gap is timestamped. Afterwards you get a scrubbable keystroke replay, a class table, a CSV, and a one-page printable authorship report per student with the figures this rubric asks for: elapsed and active minutes, characters typed versus inserted, characters deleted, the largest single insertion, paste attempts, focus losses and the longest idle gap. The report is the thing to hand a parent or print for a file.
It makes no AI-detection judgment and contains no model scoring anyone's writing — every number on the report is something the editor observed. When a session is marked, the report says in its own words that this is evidence and not a verdict. That is what makes it usable inside a rubric: observations survive being argued with, and guesses do not.
The limits, said plainly
Process criteria reward students whose process looks conventional, and not every good writer drafts messily — some plan in their head and type close to final, which scores low on Revision while being perfectly honest work. Read that as a reason to ask, not as a penalty, and let a student explain their process. Recorded sessions cover writing done in the tool, so a piece drafted at home is outside the record entirely. Timed and observed writing needs accommodation for the students who need it, and a rubric that quietly punishes slow processors is a bad rubric. And nothing here proves authorship on its own: a student can retype from a phone, and the record will say "typed", because they did. Use the evidence to start conversations, keep the marks on the writing, and be as explicit with students about the limits as you are with parents.
Record a draft and try the rubric
Want to see the report first? The public demo dashboard has one clean session and one with paste attempts and a large insertion.
Read next
- In-class writing assignments that can't be pasted — six designs that produce this evidence.
- Keystroke replay for student essays — every field on the authorship report, and what is stored.
- An AI detector flagged my student's essay — the conversation when a score comes back high.
- Google Docs version history vs a process replay — the free option, and where it stops.
- How to prove a student wrote it — evidence versus detector scores.
- Paste-proof writing assignment — running the session on Chromebooks and phones.
- The AI-proof in-class essay — how the period is structured, with an honest FAQ.