Nearly every AI tutoring product now claims to be "personalised" or "adaptive." Most of that claim is marketing. The good news: you don't need to understand any of the underlying technology to tell the difference. Three questions, tested during any free trial, will tell you which apps actually change what your child sees next โ€” and which ones just reshuffle a fixed worksheet behind a chat window.

Why this is worth five minutes before you subscribe

"Personalised" sells. Parents are, reasonably, willing to pay more for a tool that meets their child where they are rather than where a generic Year 8 scheme of work assumes they are. Product marketing knows this, which is exactly why the word gets used so loosely.

The honest test isn't the word on the landing page. It's what actually happens the moment your child gets a question wrong, in real time, in front of you.

Question one: does the difficulty change immediately after a wrong answer?

Sit with your child for their first ten minutes on any trial. Let them get a question wrong on purpose โ€” pick something they genuinely find hard, or answer carelessly.

Real personalisation: the next question is noticeably easier, or offers a hint, or breaks the same problem into a smaller step. The system reacted to what just happened.

Marketing dressed as personalisation: the next question comes from the same fixed sequence regardless of the last answer โ€” it might even be harder, because the sequence is "expanding brackets, question 4 of 10" and question 4 comes after question 3 no matter what.

This is the single fastest tell. If nothing changes after three wrong answers in a row, nothing is actually adapting.

Question two: who sets the difficulty โ€” this week's performance, or a fixed setting?

Open the settings or ask support directly: is difficulty driven by how your child is doing right now, or by a "Year 8" / "KS3 Higher" label chosen once at sign-up?

Real personalisation: difficulty tracks recent performance โ€” the last few sessions, sometimes the last few questions โ€” and moves up or down as mastery shifts. A student who's had a strong week gets pushed a bit further; a student who's struggling with a specific topic this week gets more support on that topic specifically.

Marketing dressed as personalisation: difficulty is essentially a dropdown chosen once ("Year 8," "Higher tier") that barely moves afterwards. The label creates an illusion of tailoring while the actual content stays close to identical for every child who picked the same setting.

A fixed setting isn't useless โ€” it's a reasonable starting point. The problem is when it's presented as ongoing personalisation rather than what it actually is: a one-time curriculum filter.

Question three: does it report skill mastery, or does it report a ranking?

Look at whatever progress screen the app shows a parent. Two very different things get shown here, and only one of them is educationally useful.

Real personalisation: a report that names specific skills โ€” "confident expanding single brackets, still needs practice with negative coefficients" โ€” tied to what your child has actually attempted.

Marketing dressed as personalisation: a headline number like "your child is in the top 15% nationally" or a streak counter, with little underneath explaining which specific skill that number is built from.

Percentile and ranking displays aren't wrong to include, but on their own they're an engagement lever, not a teaching signal. They're built to make a parent feel reassured (or anxious enough to keep the subscription active) โ€” not to tell you what to actually work on this week.

Put all three together

The pattern behind all three questions is the same: is the app reacting to your specific child, this week, or running a script that looks personal because of the tone of voice and the branding?

  • Difficulty that moves after a wrong answer โ†’ reacting to your child
  • Difficulty set by recent performance, not a one-time label โ†’ reacting to your child
  • Reports naming specific skills, not just a ranking โ†’ reacting to your child

If an app answers yes to all three, the "personalised" claim is probably earned. If it answers no to two or three, you're likely paying for a well-designed worksheet with a chatbot attached.

A note on why this matters more than it looks like it should

This isn't pedantry about marketing copy. Genuine step-by-step personalisation is the mechanism behind why one-to-one tutoring dramatically outperforms classroom teaching โ€” see the Bloom two sigma problem for the research behind that claim, and step size explained for exactly what "reacting to your child" means question by question. An app that fails all three questions above isn't offering a slightly weaker version of that effect โ€” it isn't offering it at all.

FAQ

How do I know if an AI tutoring app is actually personalised or just marketing?

Ask three questions during a free trial: does difficulty change immediately after a wrong answer, is difficulty driven by this week's real performance rather than a fixed year-group setting, and does it report specific skill mastery rather than a ranking. Genuine personalisation answers yes to all three.

Why do so many AI tutoring apps show a percentile or national ranking?

It's an engagement and retention tactic. "Top 12% nationally" triggers comparison and keeps a subscription feeling worthwhile, but it doesn't tell you which specific skill your child needs to practise next.

Is it worth paying more for a genuinely personalised AI tutor?

If your child uses it daily, yes โ€” the pacing is the entire mechanism behind one-to-one tutoring's advantage over a classroom. For occasional revision, a cheaper fixed-content app may be enough. Test with the three questions above during a free trial first.

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