UK teachers spend roughly 8.7 hours a week marking — and that single number explains why assessment technology is now growing faster than any other corner of the sector. The figure comes from the DfE Teachers' Workload Survey of academy schools, not from any new release; what is new is the market response. The Department for Education's June 2026 report, Assessment of the education technology market in England, doesn't restate the marking-hours figure, but it captures the consequence: assessment is the fastest-growing vertical in UK EdTech (educational technology), expanding at 18.1% a year (p.8). Schools, in short, are spending their way towards the workload relief that policy alone never delivered.
How much time do UK teachers actually spend marking?
The headline figure is about 8.7 hours a week, drawn from the DfE Teachers' Workload Survey covering academy schools. It is corroborated by the OECD's TALIS 2018 study, which recorded teachers spending around 9.4 hours a week (secondary) and 9.7 hours a week (primary) on assessing and marking pupil work. Whichever source you take, marking sits among the very largest blocks of the working week — and TALIS 2018 put that total working week at 49.3 hours, with 57% of secondary teachers describing their workload as unmanageable.
| Measure | Figure | Source |
|---|---|---|
| Marking & assessment per week | 8.7 hours | DfE Teachers' Workload Survey (academy schools) |
| Assessing/marking, secondary | 9.4 hours/week | OECD TALIS 2018 |
| Assessing/marking, primary | 9.7 hours/week | OECD TALIS 2018 |
| Total average working week | 49.3 hours | OECD TALIS 2018 |
| Teachers calling workload unmanageable | 57% (secondary) | OECD TALIS 2018 |
Why hasn't the workload problem moved in a decade?
Successive workload-reduction reviews, marking-policy guidance and "reducing teacher workload" toolkits have come and gone, yet the marking figure has stayed remarkably steady. The reason is structural. Marking time is a function of class sizes, the volume of extended writing students produce, and a long-standing professional expectation that pupils receive regular, individual written feedback. None of those drivers has fundamentally changed.
Policy can tell a school to mark less, but it cannot make the marking itself faster — and where written feedback genuinely helps learning, schools are reluctant to cut it. That tension is exactly why attention has shifted from changing the rules around marking to changing the tools that do it. If you want the wider context for why assessment, specifically, has become the centre of gravity, the first article in this series looks at why assessment is the fastest-growing UK EdTech vertical.
What does the DfE's 2026 EdTech report add to the workload conversation?
The 2026 report's contribution is not another workload statistic — it is evidence that schools are buying their way out of the problem. The assessment vertical is now the fastest-growing in UK EdTech, with 71 companies generating £470m in turnover and growing at 18.1% a year, the highest rate of any category (p.8). And the DfE is explicit about what is driving that demand.
That framing matters because workload reduction is baked into how the DfE defines the whole category. The report describes EdTech as technology "designed to improve the learning experience for pupils, reduce workload for staff, and improve the efficiency of operations" (p.13). In other words, when schools spend on assessment tools, they are spending against the marking number — the 8.7 hours that policy never managed to shift. The market signal and the workload data are two halves of the same story.
What does effective AI feedback actually look like?
Not all of that spending is equal, and the DfE flags a clear direction of travel: away from general-purpose GenAI (generative AI — systems such as ChatGPT and Claude) towards specialised tools. The report notes that "investment shifts towards specialised AI solutions, and general-purpose AI may reduce the competitive advantage of legacy platforms" (p.8). For marking, that distinction is everything. A general chatbot can produce fluent feedback, but it does not know the AQA, Edexcel, OCR or WJEC mark scheme an answer will be measured against — so its "marking" is a guess dressed up as authority.
Tools built specifically for UK curricula illustrate the shift. ReMarkAble AI, for example, was trained on AQA, Edexcel, OCR and WJEC mark schemes and returns examiner-style feedback on extended writing in under 60 seconds — including grades, strengths, areas for improvement and next steps. The point isn't the product specifically; it's that the category — curriculum-aligned AI feedback — is exactly what the DfE's data shows schools buying. The bar for "effective" here is feedback that mirrors how a real examiner reasons, not generic prose; for more on what separates the two, see examiner-quality AI marking and moderation.
What should school leaders do in the next 12 months?
The practical implication of the data is to treat assessment technology as a workload investment with a measurable return, not a gadget. A sensible 12-month approach looks like this:
- Measure the baseline. Establish how many hours your teachers actually spend marking now, by phase and department, so any tool can be judged against it.
- Favour the specialised over the general. The DfE's own signal points to curriculum-aligned tools over general-purpose chatbots for marking. Be cautious of GenAI platforms repurposed for assessment without mark-scheme grounding.
- Pilot small, formatively. Run a single class set or department trial on practice work first, where AI feedback carries no awarding risk, and compare the feedback against your own moderation.
- Keep teachers in the loop. Use AI for the fast first pass and the volume of practice feedback; keep professional judgement on final, moderated marking. Many schools are already running general GenAI alongside this — our companion piece on ChatGPT in schools and what the DfE says about GenAI covers the governance side.
The 8.7-hour number has resisted a decade of policy. The 2026 report suggests schools have stopped waiting for policy and started buying — and the smart move is to buy deliberately, against a baseline, with curriculum-aligned feedback as the bar.
Try curriculum-aligned AI marking with one class set
If your department is exploring AI marking, the no-cost route is to try a curriculum-aligned tool with a single class set. ReMarkAble AI marks against AQA, Edexcel, OCR and WJEC criteria from KS1 to A-Level, is free to use, and requires no card to start.
Frequently Asked Questions
How many hours per week do UK teachers spend on marking?
Government data has put the figure at roughly 8.7 hours a week. That number comes from the DfE Teachers' Workload Survey covering academy schools, and it is reinforced by the OECD's TALIS 2018 study, which recorded around 9.4 hours a week on assessing and marking for secondary teachers and 9.7 hours for primary teachers. Marking is consistently one of the single largest components of the working week. It is also one of the most stubborn — the figure has barely shifted across successive surveys over the past decade.
Has teacher workload in England gone down?
Not meaningfully on the marking front. Despite a decade of workload-reduction policies, marking and assessment time has stayed broadly flat. TALIS 2018 put the total average working week for secondary teachers at 49.3 hours, and 57% of secondary teachers described their workload as unmanageable. The pressures that drive marking time — class sizes, the volume of extended writing, and the expectation of regular written feedback — have not gone away, which is why schools are now turning to technology rather than policy to move the number.
Can AI mark GCSE essays accurately?
Curriculum-aligned AI marking tools can produce feedback that closely follows exam board mark schemes for structured, criteria-based answers — checking whether an answer includes the expected evidence, terminology and structure. They are strongest on essay-based subjects where there are clear levels of response, and weaker on highly creative or unconventional work. The honest framing is that AI marking is best used formatively, to guide improvement, rather than as a final summative grade. Used that way, it is reliable enough to help students and teachers see strengths and gaps quickly.
What's the difference between AI marking and AI feedback?
AI marking assigns a grade or level against a mark scheme; AI feedback explains why — pointing to specific strengths, weaknesses and next steps. The most useful tools do both, but the feedback is what actually reduces workload and improves learning, because it tells a student what to change. A general-purpose chatbot can produce plausible-sounding feedback, but it does not know the AQA, Edexcel, OCR or WJEC criteria a UK answer is measured against. Curriculum-aligned tools tie the feedback back to the specific mark scheme, which is the whole point.
Are AI marking tools approved by exam boards?
No exam board currently certifies third-party AI marking tools for awarding live grades, and any tool claiming official board endorsement should be treated with caution. What good tools do is mark against the published mark schemes and assessment criteria, so the feedback mirrors how an examiner would approach the work. For schools, this means AI marking sits firmly in the formative and practice space — helping students improve and giving teachers a faster first pass — rather than replacing the formal, moderated marking that determines final results.