AI Marking Software for Heads of English

The English department carries the heaviest marking load in any secondary school. Every essay is long-form, every mark scheme is banded rather than points-based, and the cohort spans Year 7 creative writing through to Year 13 critical essays. A head of English evaluating AI marking software for schools is doing so under different pressures from any other head of department — the workload is genuinely larger, the marking is more interpretive, and the teaching staff are more likely to be the colleagues going home with a bag of books on a Friday.

This guide is for the head of English, or the deputy or head of teaching and learning supporting them, evaluating AI marking before a September rollout. It covers what English marking actually needs from a tool, how mark-scheme matching behaves on long-form essays, how the shared school credit pool changes the budgeting conversation with SLT, how onboarding works without a procurement nightmare, and what happens to the student work that flows through the platform.

What English Marking Actually Needs

Generic AI feedback is worse than useless for English. "Good use of evidence" written across the top of a Year 11 literature essay does not move the student up a band, does not satisfy the moderator, and does not give the teacher anything to write in the report. English marking needs awards mapped to the specific assessment objectives, banded against the published level descriptors, with feedback that names the move in the writing that earned each mark.

The tool also has to handle the full English shape. Creative writing against AO5 and AO6. Critical essays against AO1 through AO4. Comparative essays where the marker is balancing two texts. Language analysis where the marker is following a quotation through to its effect. Each is a different marking shape, each needs a different chunk of the published mark scheme to be the reference, and each needs a tool that does not just produce a generic "this is a level 5 response" verdict.

GradeOrbit's marking workflow takes the actual board mark scheme as an upload — AQA's level descriptors, Edexcel's banded criteria, OCR's content grids — and marks against that document. The marks come back tied to the criteria. The feedback names the moves in the writing. A head of English evaluating the tool should spend the first hour of the trial running a piece they have already moderated, comparing the AI output to their own annotations, and deciding whether the tool's reading aligns with the department's reading.

Consistent Marking Policy Across the Department

The departmental marking policy is the head of English's signature artefact. Standardisation meetings exist precisely because Year 11 essays marked by five different teachers need to reach broadly the same band for broadly the same answer. The risk of any AI marking rollout is that it pulls marking away from the agreed standard — either by being lenient in a way the department is not, or by being strict in a way the moderator will reject.

The mitigation is calibration before rollout. The head of English picks a set of previously moderated essays at known levels, runs them through the AI, and compares the output to the agreed marks. Where the AI lands within a band of the department's standard, it is in calibration. Where it consistently sits one band high or low, the prompt or the mark scheme upload needs adjustment before any teacher uses it on live work. The tool serves the department's policy, not the other way around.

For the cross-department picture, see our guide to rolling out AI marking across your department and the post on AI marking software for heads of department. Both walk through the calibration and pilot sequence in more depth.

Shared School Credit Pool — One Budget, Every Teacher

Per-teacher subscriptions do not work in an English department of twelve. They create twelve procurement relationships, twelve usage patterns to track, and a fairness problem the moment one teacher's set is heavier than another's. The right model is a single school credit pool — funded by SLT or by the department from its own budget — that every teacher in the department draws from.

GradeOrbit's school tier is exactly this. One billing relationship, one credit pool, every teacher in the school marking against the same pool. The head of English can see departmental usage on the dashboard, raise it with the deputy head if the burn rate is higher than the agreed cap, and reallocate where needed without going back to the supplier. The cap sits with SLT. The day-to-day usage sits with the teacher. The procurement footprint is one supplier, not twelve.

Onboarding Without a Procurement Nightmare

School procurement can take a term. A head of English starting the September term with twelve teachers in the department needs the tool live in the first fortnight, not in the second half-term. The onboarding flow that works is a fast initial sign-up by the designated signatory using a school email address, the school's URN as an optional rather than mandatory field, and the procurement paperwork — Data Protection Agreement, supplier review, approved-list addition — running in parallel rather than blocking the start.

GradeOrbit's school onboarding follows that pattern. The signatory signs up, the DPA is reviewed and countersigned, the billing relationship is established, and teachers are invited in. The whole sequence is measured in days, not in a term. The DPA itself is a real document, reviewed by SLT and the school's data protection officer, not a clickthrough.

For the SLT view of the procurement review, our post on AI tools for secondary schools: senior leaders' guide covers the standard set of questions a procurement-aware SLT will want answered.

Student Data: What Never Gets Stored

English departments mark more student work than any other department. The data-handling posture of the tool matters more for English than almost anywhere else in the school, simply by volume. The wrong answer to "what happens to the essays we upload" is "they are stored on our servers and used to improve the model". The right answer, and the only one a school DPO should accept, is "nothing is stored, ever".

GradeOrbit does not save uploaded student work. Every page is processed in memory and discarded the moment the marking result is returned. Student names are redacted on the page before upload using a black-box tool that burns the redaction into the image client-side, so the AI model only ever sees the anonymised version. Students are referred to as "Student 1", "Student 2", and so on. The detail is in the DPA, not in a marketing footnote.

Talk to Us About GradeOrbit for Your School

If you are a head of English evaluating AI marking for a September rollout — or an SLT member supporting one — the next step is a conversation about how the tool fits your department. Head to the GradeOrbit homepage to see the school-tier overview and get in touch from there to discuss deployment. We can walk through the DPA, the credit-pool sizing for your department and wider school, the calibration sequence against your existing marking policy, and the rollout plan that fits your summer-term timeline.