We stopped re-marking homework

The fifteen minutes we were losing

For two years, every session at our centre started the same way. The student sat down, pulled out last week's homework, and the tutor marked it while the student waited. Fifteen minutes, sometimes twenty. Four tutors, four rooms, the same ritual.

We told ourselves it was diagnostic. It wasn't, really. By the time I was marking a worksheet, the student had done it five days earlier, possibly with help, possibly in a rush in the car on the way in. A wrong answer could mean the idea never landed, or that they'd understood it on Wednesday and lost it by Monday, or that a parent had tried to help and got it wrong. I couldn't tell which, and I was spending a quarter of a paid hour finding out that I couldn't tell.

The other cost was the one we didn't talk about. Four tutors mark four different ways. I was lenient on spelling in written answers, because my background is in language teaching and I can see what a second-language writer is reaching for. Lena, who does maths, marked method. Sam marked the final answer and nothing else. The same student, rotating through the centre over a term, was being measured on three or four different scales, and nobody had written any of them down.

Check: Why was marking last week's homework at the start of a session a poor way to find out what a student had learnt?

Why the end of the session is the right moment

What changed our minds wasn't software. It was a conversation in the kitchen about when a gap is cheapest to close.

A gap found at the end of today's session costs five minutes. The idea is still warm, the example is still on the whiteboard, and the student can be shown where it went wrong while they still remember what they were thinking. The same gap found a week later costs most of a session, because you have to rebuild the context before you can even look at the mistake.

There's a second reason, and it's the one the research is clearest on. Being asked to pull an idea back out of your head, shortly after you put it in, does more for keeping it than reading it again does. Teachers call it retrieval practice; the studies call it the testing effect. The point for a tutor is practical: a short check at the end of a session isn't only a measurement. It's part of the teaching. The student who answers five questions about today's work at 4:55 walks out having rehearsed it once more than the student who didn't.

So the check moved. Not "next week, at the start", but "today, before you go".

Check: Give two reasons a short check at the end of a session is better than checking the same material a week later.

What ten minutes actually looks like

Here is the shape we settled on, after a month of getting it wrong (more on that below).

• Five to seven questions. Never more. Ten minutes includes reading the result together.
• One idea. The thing today's session was actually about. If the lesson was the present perfect, every question is about the present perfect. Not "English".
• A mix. Three or four choice questions that target the mistakes students actually make, and one or two written answers where the student has to produce the thing rather than recognise it.
• On their own phone. They answer sitting where they are. It's marked as they submit, and they see which ones they got, with a line of explanation under each.
• We see the same screen. The result is in front of the tutor before the student has picked up their bag.

The draft of the questions is written from the tutor's own notes for that session. The tutor reads every question, changes what needs changing, and only then does the student see it. That reading takes two or three minutes. It is the most important two or three minutes in the whole routine, and it is the part I'd never give up.

Check: What are the four features of a ten-minute end-of-session check as the centre runs it?

Four tutors, one standard

The surprise benefit was consistency.

When each tutor writes their own questions by hand, you get each tutor's habits. Lena's questions were always a little too hard. Mine had a tell: the longest option was always the right one, and the sharper students had worked that out. Sam didn't write checks at all when he was busy, which was most weeks.

Now every tutor's check starts from the same place: a draft written from that day's notes, which the tutor edits. Lena still makes hers harder. But the floor is the same across the centre, and nobody skips the check because writing it from scratch was the hard part.

The marking is consistent too, and this matters more than it sounds. A written answer is marked against the answer the tutor expected, and it's the same marker for every student, every time, at 3 pm and at 7 pm. We used to have a quiet problem where a student who moved from Sam's Tuesday slot to my Thursday one suddenly "improved". They hadn't. I was the softer marker. That is gone.

What we gave up for it is honest to say: the tutor can't nudge a mark after the fact. If I think a written answer deserved more, my job is to fix the expected answer so it marks the next student properly, not to argue with this one. It took me a fortnight to stop resenting that and another fortnight to prefer it.

Check: How did the centre get consistent marking across four tutors, and what did the tutors give up for it?

What homework turned into

Once the check was doing the diagnosing, we had to ask what homework was for.

The answer turned out to be practice. Homework is where the student does twenty of something, slowly, without a tutor watching. It was never a good diagnostic, because we couldn't see how it was done. It is a very good place to get the reps in.

So we stopped marking it line by line. The student still does it. The tutor glances at it, asks what was hard, and moves on. If a student wants to know whether they're getting it, the answer is in the check from last session, and in the retake they can do at home before the next one.

That retake is the part parents noticed. A student who got three out of six on Thursday can sit the same check on Saturday, at home, and both of us see the second score next to the first. Nothing is overwritten. The pair of attempts says more than either one on its own, and it took the "did you do your homework?" conversation out of the house entirely. Nobody asks that anymore. They ask "did you have another go?"

Check: After the end-of-session check was introduced, what was homework for, and how did the centre stop marking it?

How the next session starts now

The first five minutes of a session used to be marking. Now they are the result from last time.

The tutor opens the student's last check before they arrive. There is a question or two they missed, and usually a pattern: a student who gets the form of a tense right and the use of it wrong, every time; a student who can simplify a fraction but not compare two. The session opens there. "Last week you had this one, let's look at it", and then into the new material.

On the pages we run for a single student, the next set of questions also leans toward what that student missed, so the check at the end of this session comes back to last week's gap once more before moving on. We didn't plan that. It was there, and it turned out to be what good tutors do by instinct anyway: circle back without making a thing of it.

Over a term this becomes a record that a single tutor couldn't keep by hand. Every check, every attempt, every retake, per student, in order. When a parent asks how it's going, we open it rather than remembering.

Check: What do the first five minutes of a session look like now, and where does that material come from?

What we got wrong in the first month

An honest list, because the first month was rougher than the paragraphs above suggest.

Too many questions. Our first checks were twelve to fifteen questions. Students rushed, the last few answers were noise, and the "ten minutes" was twenty. Cutting to six fixed almost everything.

Questions about the handout, not the idea. A draft written from a handout will happily ask what the third example on the page was. That tests whether the student was looking at the page. Lena's rule now: if a question could only be answered by someone who had read this specific sheet, delete it.

Expected answers that were too strict. The first written-answer questions were marked hard, because the expected answers I wrote were long and specific, and the marker held students to them. Shortening the expected answer to the idea in plain words, and letting the marking follow that, fixed it. I'll write a whole post about the edits I make, because it's where the craft is.

Reading the draft in a hurry. On a busy afternoon I approved a set without reading it properly, and one question had two defensible answers. A student spotted it, which is the best possible outcome and still embarrassing. Nobody approves without reading now. Two minutes.

One tutor who didn't do it. For three weeks Sam ran sessions without a check, because it was new and he was busy. His students' records had holes and his parents noticed, because the other three tutors' parents were getting a result every week. Peer pressure did what a meeting wouldn't have.

Check: Name three mistakes the centre made in its first month of end-of-session checks and how each was fixed.

A term later

I'm wary of numbers in a post like this, because ours are four tutors and about sixty students, and you should treat them as one centre's experience, not a study.

Marking time per session went from fifteen or twenty minutes to about three, nearly all of it reading the draft. The time between a gap appearing and a tutor seeing it went from a week to the same afternoon. Roughly a third of students do a retake at home without being asked, and that third is not the third you'd guess: it's mostly the students who got four out of six and wanted the six.

The thing I didn't expect is that sessions feel different. The last ten minutes used to be the tutor winding down and the student watching the clock. Now they're the most alert part of the hour. There's a small, friendly pressure in knowing a check is coming, and students lean into the lesson because of it.

Check: What changed in the centre's marking time and in how quickly gaps were found, and why should a reader be cautious about those numbers?

If you want to try this with no software at all

The principle matters more than the tool, so here is the paper version. It works. It's what I did for years as a classroom teacher before any of this existed.

1. In the last ten minutes, give the student five questions on an index card. Write them before the session, while you plan, not during.
2. Make three of them choice questions where the wrong options are the mistakes you've actually seen students make, and two of them "explain" or "write an example" questions.
3. Mark them together, straight away, out loud.
4. Keep one sheet per student with a row per session: date, topic, score, the one thing missed.
5. Start the next session from that row.

What you won't get on paper is the consistency across tutors, the retake at home, or the record that keeps itself. But you'll get most of the learning benefit, and you'll find out within a fortnight whether your centre wants the rest.

Check: What are the five steps of running an end-of-session check on paper?

If you're in Australia

Most centres here run on the school's four terms, about ten weeks each, with sessions once or twice a week. That rhythm suits an end-of-session check well: a ten-week term is around ten to twenty checks per student, which is enough history to see a pattern and short enough that a parent conversation at the end of term can open with the actual record rather than an impression. If your centre bills by the term, the record is also the plainest answer to "what did we get for that?". Swap this section for your own country's rhythm; nothing above depends on it.

Check: Roughly how many end-of-session checks does one ten-week term produce for a student who attends once or twice a week?

Try it with your own notes →