Adaptive Learning: Meaning, How It Works, What Indian Research Shows, and How a Coaching Institute Can Start It

What adaptive learning means, in plain words

Think of a good private tutor with one student. The tutor asks a question, watches the answer, and moves on or goes back one step. Adaptive learning tries to bring this habit to many students at the same time.

Every adaptive system, simple or advanced, keeps asking the same three questions about each student:

1. Where is this student now? A diagnostic test or the last few answers show what the student knows and does not know.
2. What should this student do next? A rule or a program picks the next lesson, practice set or level of difficulty.
3. Did it work? The next answers show whether the student has learned it. If yes, move on. If no, try another way.

The U.S. Department of Education, in its 2023 report on artificial intelligence in teaching, notes that the word personalisation is often used for adaptivity but means different things to different people. For some it means giving students voice and choice. For others it means a learning management system (LMS) that recommends a playlist of activities to each student. The same report says that adjusting the difficulty and the order of lesson materials are two of the most common ways that education software adapts. Keep this in mind when a company says its product is adaptive: ask what exactly it changes.

Adaptive learning and the words close to it

How adaptive learning works: the loop

Adaptive learning runs as a loop. Each turn is small: a few questions, one short lesson, one check.

1. Diagnose. A short test finds the student's level on each skill, not just one total score.
2. Choose. A rule picks the next activity. For example: below 50 percent on fractions, go to the fractions repair set.
3. Teach and practise. The student watches a short lesson or works on a practice set at the right level.
4. Check. A few new questions test the same skill again.
5. Adjust. Move ahead, repeat with a different explanation, or call the teacher.

An adaptive course can change five things from one student to another:

Types of adaptive learning, from simple to advanced

Adaptive learning is not one product. It is a range, from a teacher's simple rule on paper to an artificial intelligence (AI) tutor that responds to every step a student writes. Most coaching institutes should start at the top of this table and move down only when they have the data and the need.

A well-known example of a section-level adaptive test is the GRE General Test, used for admission to many graduate schools. Its official page says the Verbal Reasoning and Quantitative Reasoning measures are section-level adaptive: the first section of each is of average difficulty, and the difficulty of the second section depends on how well the test taker did in the first. A coaching institute can copy this idea for its own mock tests without any special software, as the template later in this guide shows.

What research says about adaptive learning

Adaptive learning has been studied for decades, mostly under the name intelligent tutoring systems. The results are encouraging, but they come with conditions that matter for any institute.

What does an effect size mean? It compares results across studies. In the Kulik and Fletcher review, an effect of 0.66 meant the average student using the system moved from the 50th to about the 75th percentile of the comparison group.

Three warnings from the same studies are worth more than the headline numbers:

• The test matters. Kulik and Fletcher found gains depended greatly on whether local or standardised tests were used. Gains on your own chapter test may not show up in a board exam.
• The setup matters. In the same review, evaluations with poor implementations of tutoring systems showed small effects. Buying software is not the same as using it well.
• Teachers still matter. Ma and others noted some evidence that teachers and teaching methods are important in learning with these systems.

The Indian evidence: a study in Delhi

The strongest evidence on adaptive learning in India comes from a randomised study published in the American Economic Review in 2019 by Karthik Muralidharan, Abhijeet Singh and Alejandro Ganimian. It studied after-school centres in low-income areas of Delhi that used Mindspark, adaptive software made by the Indian firm Educational Initiatives.

How the program worked

• The centres ran six days a week, 90 minutes a day: 45 minutes on the adaptive software and 45 minutes of group support from an instructor, in groups of 12 to 15 students.
• The software started with a diagnostic test and then adapted after every activity, so each student worked at their own level, not at the level of their class.
• It drew on an item bank of more than 45,000 test questions and used patterns in students' mistakes to choose remedial help.
• 619 students from public middle schools took part. About half won a lottery that gave them free attendance.

What it found

• After about 4.5 months, lottery winners scored 0.37 standard deviations higher in maths and 0.23 higher in Hindi than students who did not win.
• Gains in absolute terms were similar for all students, but much larger relative to their usual progress for weaker students.
• Students in the same class were spread across five to six grade levels. In grade 6, the average student was about 2.5 grade levels behind in maths, and by grade 9 nearly 4.5 levels behind.
• The authors write that the program raised the productivity of an instructor, not technology by itself, since an instructor was part of every session.

The lesson for a coaching institute: adaptive learning helps most when one batch holds very different levels, and it worked with a teacher in the room, not in place of one.

Teaching at the Right Level: adaptive learning without computers

The Indian NGO Pratham developed Teaching at the Right Level in the early 2000s. According to J-PAL, children are assessed with a simple tool, grouped by learning level rather than age or grade, and taught from what they already know, with ongoing simple checks of progress. J-PAL says the approach has reached over 60 million students in India and Africa. This is adaptive learning done by teachers in groups, and it is the model behind the low-cost plan in this guide.

What NEP 2020 says about adaptive learning

India's National Education Policy 2020 uses the word adaptive in several places:

NEP 2020 does not ask anyone to buy a product. Its focus is continuous assessment that changes teaching for each student.

How to start adaptive learning in a coaching institute, step by step

You do not need expensive software to start. You need a clear map of topics, a good diagnostic test, written rules, and material at more than one level. Here is a plan for one chapter. Once it works, repeat it for the next chapter.

Step 1: Break the chapter into small skills

Write down the five to eight skills the chapter needs, including older skills it depends on. Trade-off: too few skills and the test cannot tell you what is wrong; too many and the test becomes too long.

Step 2: Write a short diagnostic test

Write two or three questions for each skill, from easy to medium. Keep the whole test under 20 minutes. Each wrong answer option should match a common mistake, so a wrong choice tells you why the student went wrong. Our guide on how to make a question paper explains blueprints and good options.

Step 3: Set clear rules

Decide the cut-off scores before you see any results, and write them down. A simple rule: on each skill, 2 or 3 right out of 3 means ready; 1 right means needs practice; 0 right means needs re-teaching. Rules written in advance stop you from moving everyone ahead because the syllabus is running late.

Step 4: Prepare material at three levels

For each skill, prepare a repair set (short explanation, worked example, easy questions), a practice set (normal questions) and a challenge set (harder or mixed questions). This is the biggest piece of work. Start with the skills where most students fail, and reuse material across batches every year.

Step 5: Group students by skill, not by rank

Make two or three groups for each skill. A student can be in the challenge group for integers and the repair group for word problems. Change groups after every check, so no one gets stuck with a label.

Step 6: Check again within a week

A short re-test on the same skills, with new questions, tells you who has moved. Students who now pass move up. Students who still fail need a different explanation, not the same one again, and some of your own time in a small group.

Step 7: Look at the numbers once a week

Count how many students are in each group for each skill. If most of the batch fails one skill, the problem is probably the teaching or the question, not the students. Our guide to learning analytics shows how to read this kind of data every week.

Worked example: adaptive learning for one Class 8 maths chapter

This is an example to copy, not a report on a real institute. The chapter is Linear Equations in One Variable, for a Class 8 batch of 30 students who meet five days a week.

Day 1: the diagnostic test

The teacher gives a 12-question test, three questions each on four skills: A, integers and signs; B, simplifying expressions; C, solving equations with the variable on both sides; D, turning word problems into equations. It takes 15 minutes. The teacher marks it skill by skill, not as a total.

Day 1 evening: the rule sheet decides the groups

Days 2 to 4: three tracks in one room

Each class has three parts. For the first 15 minutes the whole batch learns the day's new idea together. For the next 30 minutes, students work on their own set from the rule sheet. During this time the teacher sits with the re-teach group only, because the other two groups can work from their sets. The last 10 minutes are a three-question exit check on the day's skill.

Day 5: the re-check

A new 12-question test, on the same four skills, with different numbers. Students who reached ready move up. Students still at zero on a skill get a different explanation the next week, and the teacher calls their parents if the same skill fails twice in a row.

What the teacher learns

Suppose 20 of the 30 students score zero on skill D, word problems. That is not 20 weak students. It means the batch was never really taught to turn words into an equation, so the teacher re-teaches it to everyone before going back to groups. Decisions come from evidence, one skill at a time.

Keep each set short. A practice set of 8 to 10 questions that a student finishes is better than a set of 40 that they abandon. The maths quizzes on Sikhami can add extra practice outside class.

Template: an adaptive mock test in two stages

Copy the GRE idea for your monthly mock tests. Strong students get a real challenge and weak students a test they can learn from.

Online, run Stage 1 as one test and send each student the right Stage 2 link. On paper, print three versions of Stage 2 and hand them out after a short marking break. See also our guide to online exam software.

Adaptive learning on an LMS, and on Sikhami

Any learning management system can support a rule-based adaptive course if it has a few features. When you choose an LMS for this, check for these:

• Question-wise and student-wise test results, not just a total score.
• Tests with a pass mark, retakes and explanations shown after each answer.
• Sections, so repair, practice and challenge material sit in separate places.
• Analytics that show who is falling behind and where students stop watching.

Sikhami is a social network for people learning in India. Coaching institutes and tuition centres get a free LMS on it: video courses with tests and certificates, classroom batches, attendance, fees and receipts. Tests are multiple choice, with an explanation for each answer, a pass mark and an optional time limit, and students can take them again. Lectures can open a set number of days after a student enrols, and course analytics show test results and where learners stop watching. Sikhami does not choose a different path for each student by itself: you write the rules, read the results, and point each student to the right section. Learn more on the free LMS page.

How to choose adaptive learning software

If you are thinking of buying an adaptive learning platform, ask these questions first. Get the answers in writing.

Student data and the law

Adaptive learning runs on student data. Under India's Digital Personal Data Protection Act, 2023, a child is anyone under eighteen. Section 9 says a business must get verifiable parental consent before processing a child's personal data, and must not do tracking or behavioural monitoring of children or target advertising at them, with exceptions the government can set. A notification of 13 November 2025 brings Section 9 into force eighteen months from that date. Read the official text or ask a lawyer before you choose a tool, and collect only data you will use for teaching.

Common mistakes, and how to avoid them

Also watch that weak students do not only ever get easy work. Make sure every student also sees grade-level work and has a clear way to climb back up.

When adaptive learning is not for you

Adaptive learning is not always worth the effort. Be honest about these cases:

• Your batch is already at a similar level, for example a crash course for students who have all finished the syllabus. A good common plan may serve them well.
• The subject is mostly open-ended, like essay writing or debate. The U.S. Department of Education report notes that AI models have mostly been better at closed tasks, such as solving a maths problem, than at open, creative ones.
• You do not have time to make material at more than one level. Start with one chapter, or with the two-stage test, rather than promising adaptive learning across a whole course.
• You are buying software to look modern. If teachers will not act on the reports, spend the money on practice tests and doubt sessions instead.

For most institutes the honest middle path is: test often, mark by skill, group by need, regroup every week, and keep the teacher in charge. That is adaptive learning, with or without software. See also remedial teaching, and browse maths topics for practice material.