Let's be blunt about this: teacher workload has been a stated DfE priority since 2016, and in that time it's got worse, not better. That's not a data problem or an awareness problem. It's a decision problem. Schools know where the hours go — they just keep avoiding the changes that would actually claw them back. This is the post that says the quiet part out loud.

Stop pretending triple-marking helps anyone

The single biggest self-inflicted workload wound in primary schools is a marking culture that the evidence stopped supporting years ago. Detailed written comments on every piece of work do not improve outcomes more than lighter-touch approaches — the Education Endowment Foundation has been clear on this. Yet schools cling to it, usually because of an unwritten cultural expectation nobody wants to be the one to challenge.

My position: if your marking policy still expects written comments on every book, that policy is actively harming your staff for no measurable pupil benefit. Live marking, whole-class feedback sheets, and structured peer assessment are not "cutting corners" — they're the evidence-based options, and the schools brave enough to switch see hours come back without any drop in progress. Change the policy. This term.

The lesson-plan theatre has to end

Here's another one schools do to themselves: detailed written lesson plans for every lesson. Ofsted stopped requiring or evaluating plan formats back in 2014. Read that again — the external pressure that created this habit has been gone for over a decade, and teachers are still writing plans nobody reads, for an inspector who doesn't want them.

If your school still expects individually written plans as proof of preparation, you are asking staff to perform planning rather than do it. Kill it. Replace it with collaborative year-group planning, a shared resource bank, and high-quality existing resources from places like Oak National Academy. Plan once, teach it for years, share it across the team.

Where AI genuinely earns its place — and where it doesn't

I'm not going to pretend AI fixes everything, because it doesn't. But on the specific problem of administrative writing — the newsletters, the standard letters home, the policy updates, the report comment banks — it is the most immediate, practical lever available to a primary school in 2026, full stop. Schools doing this properly report 2–5 hours per teacher per week returned. For a 15-teacher school that's up to 75 staff hours a week. That is not a marginal gain; it's a part-time member of staff's worth of time, every week, for the cost of some training.

What AI should not touch: anything involving a child's personal data, and anything requiring your professional judgement about a specific pupil. Draw that line hard and everything else is fair game. If you want the detail on where the line sits, our guide to AI and GDPR in schools and our safeguarding guidance lay it out.

The five moves SLT should make this term

Enough theory. If you lead a primary school, here is what I'd actually do this term, in order:

  1. Run a real workload audit — not a survey you file, an honest look at your top three time drains. Most schools already know the answers; write them down.
  2. Rewrite the marking policy — remove any expectation of written comments on every piece. Do it now, not at the next policy review cycle.
  3. Ban recreated-from-scratch planning — mandate a shared bank and year-group collaboration.
  4. Adopt one AI tool for one task — newsletters are the easiest win. Prove the time saving, then expand. Don't roll out five tools at once.
  5. Get your AI governance sorted before, not after — staff are almost certainly already using AI. A policy that arrives after the fact is a policy that's already been ignored.

The thing nobody wants to hear

Workload reduction fails in most schools for one reason: leadership treats it as something to manage rather than something to reduce. Managing workload means sympathetic emails and the occasional wellbeing day. Reducing it means changing policies that senior leaders wrote, challenging cultural habits that feel safe, and spending a bit of budget on training. The schools that actually move the needle are the ones where a headteacher decides workload is a strategic priority worth spending political capital on. Everything else is theatre.

Where AskColin fits: The tools-and-training half of this — getting AI adopted safely and making it actually stick past the first term — is exactly what AskColin does, starting with a free site visit that audits where your hours really go. The policy and culture half is on you, but you don't have to do the AI part alone. Request a free visit →

The short version

  • Detailed marking and written lesson plans are self-inflicted — the evidence and Ofsted both stopped asking for them years ago.
  • AI's real, immediate value is administrative writing — 2–5 hours per teacher per week, done safely.
  • Workload only falls when leaders choose to reduce it, not merely manage it.