What separates real CBSE school software from a generic ERP with a "CBSE template"?

Three giveaways. First, where does the HPC live: is it generated from the marks register, or does it sit in a separate tab nobody updates? Second, does the system know about the best-2-of-3 periodic test rule, or do coordinators still maintain a spreadsheet to drop the lowest score? Third, does it model streams in XI–XII, including section-level stream splits, or does it require you to create parallel pseudo-classes to make it work?

How important is data isolation when picking a vendor?

More important than it sounds, especially under the DPDP Act. Each school’s data should live in its own sealed store, file storage included, with a clear answer to "does our data ever cross the boundary for analytics or AI training?". The vendors that decline to answer this in writing are the vendors you should not pick.

Is AI grading safe for a parent dispute?

Only if the audit trail records both the AI suggestion and the teacher decision per question, per student. Vendors that auto-publish AI marks are creating a defensibility problem for the school. The teacher should remain the author of record on every signed mark.

How long does CBSE-specific onboarding take, realistically?

Two to three weeks for a school of 600–1,500 students if the vendor handles the roster import for you. Longer if the vendor expects the school to learn their workbook format. A common-sense check during the trial: hand the vendor a real (anonymised) IX–X roster and see what they hand back. The good ones return a clean import; the rest send you a list of conversion homework.

Best CBSE school software in 2026: what to look for

We are obviously not a neutral party on this question. Relearns is one of the products in the category. We have still tried to make this guide useful to a school that has not made up its mind, because the worst outcome for everyone is a school buying software that fights its workflow for the next three years. The list below is what we would test if we were sitting on the other side of the table.

What actually distinguishes a CBSE-built product from a CBSE template

Most school software in the Indian market began life as a generic education ERP (fee collection, attendance, admissions) and added a CBSE template years later. The result is often a tab labelled "Holistic Progress Card" sitting beside a tab labelled "Marks Card" with no connection between them. During a demo, this looks fine. In daily use, it doubles the coordinator’s workload because the two tabs do not share data.

The diagnostic questions to ask during a demo:

HPC integrity. Is the HPC generated from the marks register, or kept separately? Ask the vendor to demonstrate a change to a Periodic Test mark in front of you and verify the HPC view updates without a second action.
Best-2-of-3 rule. Does the system compute the best 2 of 3 periodic tests automatically? Or does someone in the school office still maintain a spreadsheet to drop the lowest score? A real CBSE product does the maths for you.
Stream and section splits in XI–XII. Ask the vendor to show a class where Section A is Science and Section B is Commerce, with the timetable working for both. If the answer is "create two separate classes called 11-A- Science and 11-A-Commerce", the data model is not actually CBSE-aware.
Co-scholastic on its own scale. Co-scholastic areas use the 5-point A–E scale and do not affect CGPA. Ask to see Art Education, Health & Physical Education, Work Education and Discipline graded; verify that they do not accidentally fold into the scholastic average. The most common DIY HPC spreadsheet mistake is exactly this.

AI features: the questions worth asking

Almost every CBSE software vendor now mentions AI. Some of it is real; some of it is a single OpenAI call wrapped in marketing. The questions that separate the two:

Audit trail. For an AI-graded paper, can you produce both the AI suggestion and the teacher decision for any question, for any student? If the AI marks are signed off in bulk without the teacher reading each one, the school has a defensibility problem with parent disputes.
Editability. Is every AI output editable before publish? A vendor whose teaching material cannot be edited is a vendor selling content, not software.
Cost visibility. Does the school see the rupee cost of AI runs at the admin level? A vendor that hides AI costs in a per-seat licence is either subsidising them temporarily or planning to raise the price later.
Fallbacks. If AI generation fails, what happens? A real product offers a manual upload at every step. A product that locks you into AI will eventually be a problem.

Data isolation and the DPDP Act

The Digital Personal Data Protection Act, 2023 has changed the conversation about school data. Schools handle minors’ personal data, and the responsibilities flow back to whoever the school chose as a processor. A school evaluating a vendor should be able to get written answers to:

Where is the school’s data stored, in which cloud region?
How is one school’s data separated from another’s, technically?
Does the vendor use the school’s data for analytics, benchmarking or AI training?
What happens to the data if the school leaves the platform?

Isolation comes in two flavours. Some products mix every school’s data in the same tables and separate it with a school-ID column; the isolation is logical, enforced by application code. Others give each school its own sealed store; the isolation is structural. Either can be done correctly, but the second is far harder to get wrong and far easier to explain to a parent. Ask the vendor which one they run, in writing.

Roster import: the realism check

The demo is always smooth because the demo data is curated. The realism check is the roster import. During the trial, hand the vendor an anonymised IX–X roster from your school: three sections, your actual subject list, the co-teachers you actually have, the parent contacts in whatever format your previous system exported. The vendors who can hand you back a clean imported workspace inside a week are the vendors who actually have CBSE-specific import tooling. The rest will hand you a list of cleanup tasks for your school office, which is the wrong direction for a relationship to start in.

What to deprioritise

The features schools think they need but rarely use. Fancy admissions workflows, deep biometric attendance integrations, in-app fee collection. These are real needs at some schools, but a single-vendor "all-in-one" product that does each of them at 70 percent quality is usually worse than the academic surface at 100 percent quality and a separate, focused vendor for fee. Pick the academic vendor for academic depth; let operations vendors do operations.

The shortlist questions, in order

The five questions we would ask, in this order, on any demo call:

Show me one student. Change one Periodic Test mark in front of me. Show me the HPC view updating without a second click.
Show me an XI class where Section A is Science and Section B is Commerce. Same timetable view.
Show me an AI-graded paper. Open the audit trail for one question. Show me the AI suggestion and the teacher decision.
Tell me how my school’s data is isolated from every other school’s data.
Hand me a roster CSV. Hand me back an imported workspace in a week.

Any vendor — including us — that does not have clean answers to all five is not ready to be your school’s system of record. We hope you ask us those five questions too.

See how Relearns answers each →

Best CBSE school software in 2026: a practical buyer’s guide.