Step size is how much harder the next question is than the one a student just answered. Every other feature of an adaptive learning tool โ€” animations, badges, a chatty AI persona โ€” is decoration by comparison. Get step size right and a Year 8 student who "hates maths" will sit through twenty minutes without being nagged. Get it wrong and the same student pushes the tablet away in under three.

A story every parent of a KS3 child will recognise

Plenty of parents have tried handing a child a colourful maths app expecting it to buy them twenty quiet minutes. Sometimes it works. Often it doesn't โ€” and the reason is rarely the graphics.

Watch closely and a pattern shows up: difficulty jumps around. One question is trivially easy โ€” solved and forgotten in ten seconds, met with a shrug and a "this is boring." A few questions later, the app jumps three grades ahead of where the child actually is. Three wrong answers in a row later, eyes are wet and the device gets shoved aside: "I'm rubbish at this."

Neither extreme is really about maths ability. It's a step-size problem. The gap between "what she just did" and "what's asked next" was either too small to matter or too large to bridge.

The fix, and why it isn't about making things harder

The fix a lot of parents (and products) reach for is "make it more challenging" or, in the other direction, "make it easier so they don't get upset." Both miss the actual lever.

The lever isn't difficulty in the abstract. It's the size of the jump from the last thing the student got right to the next thing they're asked to do. Every question should feel like standing on tiptoe to reach the next rung โ€” not a rung already at eye level, and not one a full arm's length above their head.

This is what psychologists call desirable difficulty, and it connects to something more important than test scores: self-efficacy โ€” the belief "I can actually do this." A student who keeps landing on rungs just slightly out of comfortable reach, and keeps making it, builds that belief with every question. A student who's either coasting or drowning never gets the chance.

What this looks like in KS3 maths, concretely

Take expanding brackets โ€” a topic that trips up a lot of Year 8 and Year 9 students.

A student who has just correctly expanded 3(x + 4) is ready for a small step up: maybe 2(x + 5) - 3, adding one extra operation. They are not ready for (2x + 3)(x - 4) โ€” a full jump to double brackets โ€” even though it's "the same topic" on a scheme of work. That jump might be the right size for a student who's already handling single brackets confidently and has just started making small errors on double brackets, but it's the wrong size for the student three questions into the topic.

A tutor without step-size awareness treats both students identically because they're both "on expanding brackets." A tutor with it asks: what did this specific student just do, and how confidently did they do it?

How aitutors.me implements this

Professor Pi, the KS3 maths tutor at aitutors.me, checks two things after every problem: was the answer correct, and how much hinting did it take to get there via the four-level hint ladder. A first-attempt correct answer nudges the next problem up a notch. An answer that needed the second or third hint holds the next problem at the same level, or steps it back slightly, so the gap never opens wide enough to trigger the "I'm rubbish at this" spiral.

This is also why the hint ladder itself never simply states the answer โ€” see why tutors don't give answers for the reasoning. Handing over the answer would collapse the step-size system entirely: there'd be nothing left to measure "what the student can actually do" against.

What good step size is NOT

  • โŒ It's not "let the child pick whatever they find fun." Step size still moves through the required KS3 curriculum โ€” it paces the journey, it doesn't reroute it.
  • โŒ It's not gentler-is-always-better. Too easy is just as damaging to engagement as too hard, and easier is often the direction an app tunes toward to avoid complaints.
  • โŒ It's not the same as an age-based or year-group difficulty setting. "Year 8 level" describes a curriculum stage, not what this particular student mastered ten minutes ago.

How to spot it (or its absence) at home

Watch what happens the moment your child gets something wrong on any learning app. If the next question is noticeably easier, and a run of correct answers is followed by something noticeably harder, step size is working. If the next question looks like it came from the same fixed list regardless of how the last one went, it isn't adaptive โ€” it's a worksheet with a chat window bolted on.

FAQ

What is 'step size' in adaptive learning?

How much harder the next question is compared to the one the student just answered. A well-tuned step size feels just hard enough โ€” get it wrong in either direction and engagement collapses.

Why does my child quit an app after two minutes on some days and stay on it for twenty on others?

Almost always a step-size problem: either every question is too easy and they get bored, or difficulty jumps around unpredictably and a run of wrong answers kills their confidence.

Is adjusting step size the same as letting a child learn only what interests them?

No. Step size governs the difficulty of curriculum-aligned questions, not which topics get covered. It still works through required KS3 content โ€” just paced to what the student has actually shown they can do.

Duke Harewood ยท founder, aitutors.me ยท Updated 11 Jul 2026.