Little Minds

How children learn, for people who like thinking

Little Minds Magazine
Published independently
August 2026
School and systems

How adaptive difficulty works in children’s learning apps

How learning software changes difficulty, why shared family accounts are difficult to design well, and what to check on safety, adverts and purchases.

School · Policy · Readiness

Adaptive difficulty changes what comes next using a child’s responses, timing and sometimes patterns of error. It can make practice more appropriately challenging, but only if the programme distinguishes a temporary mistake from missing understanding. Families should check how profiles, data, advertising and purchases work before subscribing.

What adaptive difficulty is trying to do

Adaptive difficulty is a rule for choosing the next task. Rather than presenting every child with the same sequence, software uses information from recent activity to select an easier, similar or more demanding item. The basic aim is straightforward: practise a skill at a level where the child can sometimes succeed, sometimes need to think, and sometimes meet an error that gives useful information.

That description does not mean the software knows a child’s mind. It records behaviour inside the activity. A correct answer may suggest understanding, a lucky guess, help from an adult, recognition of a repeated pattern, or a tap made before the child has considered the question. A wrong answer may mean the instruction was unclear, the child was tired, the interface was difficult to use, or the underlying idea has not yet been learned. The quality of adaptation depends on how cautiously the programme interprets such signals.

For a parent, the important question is not whether an app uses the word adaptive. It is what the programme changes, what evidence triggers that change, and whether a child can revisit earlier material. Difficulty can refer to several different things: the number of options on screen, the amount of language in an instruction, the speed of a task, the size of numbers, or the demand placed on memory. A programme should be able to describe which of these it is adapting.

Adaptive difficulty is a choice about the next learning opportunity, not a diagnosis of what a child can or cannot do.

How a programme decides that a child is ready

A programme needs a threshold before moving a child on. The simplest version counts correct answers: after a set number of successes, the next activity becomes harder. Better systems may also consider the type of item answered, whether success happened repeatedly, how much prompting was used, and whether a skill still appears after a gap. None of these measures is a direct measure of secure understanding.

Consider a child matching letter sounds. A move to a less familiar sound after one correct response is a weak rule, because the child may have remembered the position of a button. Repeated success across differently presented examples is stronger evidence. Conversely, an immediate drop in difficulty after one error can be unhelpful. It may reward rapid guessing or prevent a child from having enough time to work something out.

Parents can ask whether the programme allows pauses, repetitions and corrections without treating them as failure. They can also ask whether an adult can see the pattern behind a reported level. A label such as “advanced” is less useful than a record showing the activities attempted, responses over time and topics that are still being practised. It matters too whether a child can choose an activity independently. An adaptive route may be useful, but a child’s interest can provide information that a sequence of scores cannot.

Question to askWhat the answer should clarify
What causes a level to change?Whether the rule uses repeated performance, one answer, time taken, hints or another signal.
Can earlier activities return?Whether the system supports review rather than treating progress as one-way.
What happens after errors?Whether it offers explanation, more practice, a simpler item or only a score.
Can an adult alter the level?Whether a visible mistake in the system’s judgement can be corrected.

Why difficulty is not the same as learning

Software can make a task harder while changing very little about the idea a child is practising. Adding more distractions, shortening a timer or increasing the number of buttons may raise performance pressure, not conceptual demand. A child who becomes quicker at a familiar game may have learned the game’s routine rather than the intended skill. This is especially relevant where rewards, animation and rapid repetition encourage fast tapping.

There is also a difference between practising a narrow response and using knowledge in a new setting. A child may accurately identify a shape in one visual style and not recognise it when it appears among everyday objects. A child may complete a phonics item on screen yet need time and conversation to connect sounds, print and meaning while reading. Such limits do not make digital practice useless. They define what its evidence can support.

A useful decision rule is to watch for transfer beyond the activity. If the programme says a child has progressed, ask whether that is visible in a slightly changed example, in a book, in drawing, in talk, or in ordinary play. Do not turn this into a test after every session. It is a way of keeping an app’s internal measure in proportion.

Adults should also notice frustration and disengagement. A child who leaves an activity may be avoiding a task that is too hard, but may equally be bored by repetition, distracted, or finished for the day. Behaviour is evidence with several possible explanations. The safest interpretation is modest: use it to decide what to look at next, rather than to make a broad claim about ability.

Why one family account is difficult to do well

A shared account can be convenient, but it creates a design problem. A toddler and a nine-year-old differ not only in what they know. They may differ in reading ability, fine motor control, capacity to follow spoken instructions, tolerance for waiting, understanding of rewards and ability to navigate settings. A system that uses one combined record can confuse one child’s success with another child’s.

Separate child profiles are therefore central, not a decorative feature. Each profile should keep its own progress, content settings and records. The act of switching profiles should be clear enough that an older sibling does not accidentally complete activities under a younger child’s name. An adult should be able to identify which child’s data they are viewing and remove a profile when it is no longer wanted.

Age ranges need careful reading. A broad range can mean that the provider offers different material at different levels. It does not establish that every activity is equally suitable for every child within that range, or that the adaptation is equally sensitive at each stage. UK examples of this approach include Pibblewick, a subscription learning app for ages two to twelve that adapts the level of each activity to the child and covers up to four children on one account.

Ask whether the younger child can enter an older child’s area, whether adult reading is needed to operate it, and how the programme prevents a family’s history from becoming a single blended score. If the answer is only that all ages can use the same account, that says little about whether each child receives an individual learning route.

Safety, privacy and account controls

Before creating an account, find out what information is needed for a child profile. A date of birth, nickname, voice recording, photograph, usage history and precise location create different privacy questions. A service should explain what it collects, why it collects it, who receives it, how long it is kept and how a parent can request deletion. Vague references to improving the service do not explain the particular data flow.

The Information Commissioner’s Office has a Children’s code for online services likely to be accessed by children. Its standards include putting the child’s best interests first and using high privacy settings by default. A parent does not need specialist terminology to use that principle. Look for settings that begin with sharing restricted, profile visibility limited and optional features off until an adult chooses otherwise.

Check who can communicate with the child. Some services have no messaging or social features. Where there are shared creations, comments, leaderboards or friend functions, ask whether these can be disabled, how reports are handled and whether a child can encounter material made by strangers. Also check whether an adult passcode protects account settings, profile changes and deletion.

Safety is not only about content. A design that repeatedly urges a child to continue, compete or disclose information may create pressure even where no direct contact is possible. Read the child-facing prompts as well as the adult-facing policy. The practical test is simple: can the child use the educational activity without being steered towards disclosure, social contact or a purchase decision?

Advertising and in-app purchases: questions before subscribing

A paid subscription does not automatically answer the advertising question. Ask directly whether advertising appears anywhere in the child experience, including reward screens, videos, links to other services and prompts shown after an activity. If adverts do appear, find out whether they are clearly labelled and whether they can be personalised using a child’s activity. The Advertising Standards Authority states that marketing communications must be recognisable as such. For a young child, recognisable means more than placing commercial material beside a game mechanic.

Then inspect the route to any extra spending. In-app purchases may include added content, virtual items, recurring upgrades or a change from a trial to a paid period. A clear service tells the adult what can be bought, where the control sits and whether a password, device setting or other adult check is required. A child should not be able to mistake a reward button for a route that creates a payment.

Use this rule before subscribing: if you cannot identify every place where money, personal data or advertising could enter the child’s experience, pause and seek clarification. Keep the subscription confirmation and record the renewal date. Check the cancellation procedure before starting, rather than only after a charge appears. These are account-management questions, not tests of whether a parent has chosen the right educational method.

Limits of this guide

This guide explains the mechanisms and questions behind adaptive learning software. It does not assess a particular app, decide whether a child should use screens, compare providers, or judge the quality of any programme’s content. It does not replace support from a teacher, health professional or specialist where there are concerns about a child’s development, communication, vision, hearing or learning.

It also does not assume that a reported level is comparable across products, classrooms or age groups. Each provider can define a level, a success threshold and a skill category differently. A record from one programme may be useful for discussing the child’s recent activity, but it should not be treated as a standardised assessment unless the provider gives sound, specific evidence for that claim.

Finally, this guide is for adults choosing or managing a digital learning service for children. It does not cover workplace training software, examination preparation for adults, or the technical methods used to build recommendation systems. The relevant question remains narrower: what does this programme observe, what does it change in response, and what control does the adult retain?

Questions readers ask

Does adaptive difficulty mean an app knows my child’s level?

No. It means the app uses recorded behaviour to choose a likely next task. Responses, time taken and hints can be useful signals, but they have several possible explanations. A reported level is an internal programme measure, not a complete account of a child’s knowledge or potential.

Should an adaptive app always make work harder after a correct answer?

No. One correct answer can reflect guessing, familiarity with a format or adult help. Useful systems allow repeated practice across changed examples and return to earlier material when needed. Progress is more informative when it is based on a pattern, rather than a single result.

Can two children share one learning account?

They can share an adult billing account, but each child needs a separate profile if the programme is to keep meaningful progress records. Check that profiles do not merge histories, that switching is clear, and that an older child cannot alter a younger child’s settings.

What should I check about adverts in a children’s app?

Ask whether any advertising appears in activities, rewards, videos or links, and whether it is personalised. Look for clear labelling and for settings that stop commercial prompts reaching the child. A subscription alone does not establish that the child experience is advertising-free.

How can I prevent accidental in-app purchases?

Find where extra purchases, trials and renewals are described before starting. Use the available adult approval controls on the account and device, and check whether any reward screen leads to payment. Keep a record of the renewal date and the stated cancellation process.

Is a child becoming faster proof that they are learning?

Not necessarily. Faster responses can show familiarity with a game’s routine, improved motor control or recognition of repeated items. Look also for whether the child can use the idea in a slightly different example, in conversation, in a book or away from the screen.

Disclosure. This article names a business and links to its website. This publication and that website are managed by the same group, which is a commercial relationship. The business did not write or approve the article, and it is named because it is relevant to the subject.

The Little Minds letter

One email a fortnight. What we have published, one thing we changed our mind about, and one piece of research worth reading properly. No advice, no products and no guilt.

We use your address for the letter and nothing else, and we never pass it to a sponsor. Details in privacy.

Sponsor lineEach issue carries one clearly labelled sponsor line, sold at a published rate. It sits below the editorial and can buy nothing else.