← Blog

Teacher workload in India: the best-documented pain is data entry, not teaching

We graded every source we could find on teacher workload in India. The best-evidenced pains are administrative; the learning crisis sits downstream.

Evidence and data team at Circin · · 14 min read

A government school teacher in Maharashtra (her name was changed by the paper) told Careers360 last December: "For many teachers, the day now begins not with morning assembly but with login screens." We went looking for how typical that is: how much of a teacher's workload in India is teaching, and what the evidence says the rest is.

Each claim below carries a grade. A: government data, peer-reviewed work, or several independent investigations with named people. B: one credible study, or reporting with named sources. Where we could not find a source good enough to grade, the claim is not here; a few we dropped are listed near the end. The sources are at the end.

What the evidence says, in one table

Pain Grade Best figure Source
Non-teaching duty A 39–41% of the school day spent teaching (200 Delhi teachers) Accountability Initiative, 2019
Portal data entry A Teachers in seven states logging into several apps daily; the "one single app" request Careers360, Dec 2025
Punitive attendance apps B 27 MP teachers in the High Court; pay docked for late logins in Bihar LiveLaw, Scroll
Vacancies A 9,82,662 of 69,85,760 posts vacant; 1,00,843 single-teacher schools 368th Standing Committee report; UDISE+ 2025-26
Learning levels in the room A 51.2% of rural Std V cannot read a Std II text; Grade 9 maths 37% ASER 2024; PARAKH 2024
Compliance re-keying (office) A The same child entered in five systems under five deadlines MoE D.O. 14 Jul 2026; CBSE REGN/2025-26

Teacher workload in India: what takes the time, graded

Picture, say, a Class 7 teacher in a two-teacher government primary school in the last week of September. Before first period she has photographed herself for the attendance app, opened the mid-day-meal portal, and been told by the block office that the school's UDISE+ progression is due Tuesday. She has thirty-eight children across three grades in one room. This is what the evidence says about her week.

1. Conscripted into non-teaching duty (A)

Section 27 of the Right to Education Act permits teachers to be deployed for exactly three non-educational purposes: the decennial census, disaster relief and elections. In practice the list is longer. In October 2024 nearly two lakh of Assam's 2.13 lakh government teachers left official WhatsApp groups in protest at t-shirt distribution, sapling planting and cyclothon duty, as ThePrint reported. The Azim Premji Foundation's field study of 619 schools and 2,861 teachers across six states (2016 fieldwork, non-random sample) found 18.9% of teachers absent from school on the day of an unannounced visit — and only 2.5 percentage points of that was absence without reason; the rest was sanctioned leave, official duty and training. Accountability Initiative's time-use study of 200 Delhi teachers, reported in 2019, found 39–41% of the school day actually spent teaching. That sample was chosen to over-represent the busiest teachers, and the authors say so; it is the best figure that exists.

2. Many portals, one phone (A)

Careers360's December 2025 reporting, with teachers from seven states and mentions of ten states and a Union Territory, describes them "logging data into multiple government apps and portals — from attendance and mid-day meal records to infrastructure surveys and learning outcome dashboards"; one teacher in Uttar Pradesh used her own mobile data because the school Wi-Fi did not work. Chaitanya Suresh N, a government school teacher in Anantapur, Andhra Pradesh, said: "Some days I stay back for an hour just to update everything." He also gave the most precise product specification in everything we read:

"I wish there was one single app that could handle all kinds of data — attendance, enrolment, performance — instead of switching between five different ones."

Every previous attempt to grant that wish has arrived as one more app to log into.

3. Attendance apps that penalise (B)

Photo-and-location attendance on a phone, mandatory, on infrastructure that fails. Bihar's e-Shikshakosh app requires a photograph from the school campus within a morning window, and teachers told Scroll in November 2024 how it fails them — connectivity, wrong locations, a rule that repeated late logins cost a day's pay. In Madhya Pradesh, 27 teachers challenged the "Hamare Shikshak" app in the High Court in October 2025, citing network failures, facial-recognition errors and phones that could not run it; the Free Press Journal reported over a thousand guest teachers waiting on honorarium held back over e-attendance. In Punjab, in April 2026, the Democratic Teachers Front demanded relief from the MStar app after server errors marked present teachers absent (The Tribune). What these accounts share is not an objection to being counted. It is an objection to being penalised by a server that was down, with nobody to appeal to.

4. The teacher who isn't there (A)

9,82,662 of 69,85,760 sanctioned teaching posts were vacant in 2024-25, per the 368th report of the Parliamentary Standing Committee on Education tabled in August 2025. UDISE+ 2025-26 counts 1,00,843 single-teacher schools. ASER 2024 finds two-thirds of Std I–II classrooms in rural government primary schools are multigrade. The national pupil–teacher ratio looks healthy because it averages a lakh of single-teacher schools against large urban sections.

5. The room she walks into (A)

This is the fifth item, and it is a fact about the class, not about her hours. PARAKH 2024, NCERT's national survey of about 21 lakh students: Grade 6 students answered only 29% of the questions on commonly used fractions correctly; Grade 9 mathematics averages 37%, the lowest point in the system. ASER 2024: 51.2% of rural Std V children, across all school types, cannot read a Std II text. In that room our Class 7 teacher is held to syllabus completion, foundational-literacy targets and board preparation at the same time.

6. The technology teachers chose, and the technology they were given (B)

In a two-year study of government schools in two Madhya Pradesh districts, Central Square Foundation and TISS found that 50% of smart classrooms had become non-functional within two years of setup — power, device security, broken hardware, no maintenance. The same foundation surveyed 15,000 parents, children and teachers across ten states in 2025. "87% of teachers report using EdTech, with more than half of them using it daily." "96% feel that using EdTech helps them save time." "Over half (51%) of EdTech-using teachers already use GenAI, mainly for lesson preparation and delivery." 84% use at least one of YouTube, WhatsApp and Google Search; 21% use DIKSHA. The teachers found the tools themselves, on their own phones.

What the students bring into that room

Three findings with A-grade data, because the teacher's workload is shaped by them:

  • In rural India roughly three in four Std III children cannot read a Std II text. ASER 2024 puts the share who can at 23.4% in government schools and 27.1% across all schools; outlets routinely print the first number without the label. ASER is rural, and nobody measures urban budget-private schools the same way.
  • NCERT's 2022 survey of 3,79,842 students in Classes 6–12 found 81% naming studies, exams and results as a cause of anxiety. NCRB's Accidental Deaths and Suicides in India 2024 records 14,488 student suicides, up 4.3% on the year. Read NCRB as a floor: it records one cause per death and is known to under-count.
  • ASER 2024 reports rural household smartphone ownership rising from 36% in 2018 to 84% in 2024. Of 14–16-year-olds who could use one, 57% had used it for something educational in the previous week; 76% had used it for social media.

And one fact with a deadline attached. Since 2024-25, half of the marks in CBSE's year-end papers for Classes IX–XII come from competency-focused questions — case-based, source-based, applied — under circular Acad-30/2024 of 3 April 2024 (IX–X were already at 50%; XI–XII rose from 40%). ASER 2023 shows what that tests: about 85% of 14–18-year-olds could measure a length with a scale from the zero mark, and 39% when the object was placed away from zero. Generating practice of that kind, at the right level, for a class where half the children are behind, is work the teacher above has not been given time or material for. How many schools can do it today is something nobody has measured; we are trying to, in our pilot schools.

Principal and office workload: the same child entered five times

Say the headmaster of the same school, the last Monday of September. UDISE+ is due Wednesday; the CBSE registration window opens the week after; the exam-cell teacher has a list of children whose APAAR "did not generate". For principals, exam cells and front offices in government and private schools alike, the dominant complaint is the same: compliance data entry into government portals that do not talk to each other, under hard deadlines, done by one or two people.

1. One child, five systems (A)

A school enters the same child into UDISE+, its student module SDMS, APAAR, the CBSE registration and the CBSE list of candidates, plus OASIS for the school itself, under separate deadlines with no shared identity between them. The Ministry's July 2026 letter asks for UDISE+ to be complete by 30 September; CBSE's registration circular closed corrections on 28 November last year, after which "NO CORRECTION WILL BE ALLOWED BY THE BOARD". Our practical guide to this September's version, with those dates, is the UDISE+ 2026-27 calendar.

2. APAAR: blocked by a spelling (A)

The academic ID is generated by UDISE+ against the child's PEN and only when the name matches Aadhaar. CBSE's own registration circular lists why generation failed for schools in 2025: portal integration, school-versus-Aadhaar data, slow corrections, and parental consent — and the Odisha High Court's direction that the authorities add a refusal option to the consent form was made pan-India by the Supreme Court in July 2026.

3. Circulars at seven a month (A)

By our count of the CBSE Academic circulars page, the academic series ran from Acad-01 to Acad-57 between January and August 2026, with separate Skill, Examination and Affiliation series on other pages. A single-principal school has nobody to delegate the reading to. We looked for a product that reads that page and tells a school which circulars apply to it; we did not find one, and would like to hear of one.

4. A data fiduciary with no data officer (A)

Under the Digital Personal Data Protection Act, 2023, a school is a data fiduciary. The Act's schedule sets penalties of up to ₹250 crore for failing to take reasonable security safeguards. The same school is required to push Aadhaar-linked child data into government portals, usually with no data-protection officer and no IT staff. The obligations (consent records, notices, retention, agreements with vendors) are in the Act; we found nothing written for schools about how to meet them.

Also documented, less well: RTE reimbursement arrears for budget private schools, reported state by state; the document stack for recognition renewal; and the Holistic Progress Card, whose specification lists several respondent types and domains per child, per term. That last one is B: the specification is public; nobody has measured the hours.

What we dropped

The figure that "teachers spend 19.1% of school hours teaching" is everywhere and traces to no primary document; the version of the same study published by NITI Aayog gives a different sample and a different number. We use the 39–41% Delhi time-use figure instead. ASER's Std III reading figure has to be labelled government-school (23.4%) or all-schools (27.1%); we had it unlabelled in our first draft too.

One claim to strike from every deck, including ours

"Indian schools suffer from ERP fragmentation." We looked specifically for independent evidence of school-ERP failure — adoption, migration pain, lock-in, switching costs. Every result was a vendor's content marketing selling the remedy. No independent journalism, no academic study, no practitioner forum. The version that is documented six ways is the one above: schools re-key the same child into five government systems under five deadlines with no shared identity. Say that instead.

What this means for anyone building for schools

We are one of those companies, so here is what we hold ourselves to because of this reading, stated as carefully as the numbers above. A child is entered once, and the lists a school needs come out of that one record rather than being re-typed for each portal — that is the whole argument of the companion guide, and it is what Circin does today. An attendance record a teacher can contest, and a record that survives a server outage, are things we think any school software now has to have; they are on our roadmap, not yet in the product, and we would rather say so than imply otherwise.

What no source could answer

These are asserted in most edtech pitches. We could not find evidence for any of them, and each could be settled with twenty interviews.

  • Fee-default rates in Indian private schools; every figure we traced led to a fee-software marketing page.
  • How much time daily substitution and timetable rework take.
  • The working conditions of non-teaching staff — clerks, front office, lab assistants, wardens — about whom only job boards exist.
  • Teacher time-use in India at any credible scale, which is the biggest gap under this whole post.
  • Learning levels in urban private schools; what is "known" comes from ASER's rural sample.
  • The hours a Holistic Progress Card costs per student.

We are measuring the first two in our pilot schools this term and will publish the numbers, and the method, whichever way they come out.

Sources

If you have a source that contradicts a figure here, write to us; corrections are published with a dated note.

Tags: research, teachers, compliance, evidence

Research and drafting AI-assisted; reviewed by Circin Research. Corrections: support@circin.ai.

Related