The real Ed tech opportunity

 

There are as many recent articles on the fall of Byju’s – the world’s most valued ed tech startup, as they were on the rise of Byju earlier. I have commented on the deluded thinking around unicorns in my earlier blog posts.https://rpdeans.blogspot.com/2023/08/the-coming-unicorn-meltdown.html

While Byju’s had problems with corporate governance, ethical concerns around its selling, unrealistic (in retrospect) expectations around growth from investors – as do other startups, I want to talk about opportunities I believe Ed-tech startups are missing and how Ed-tech, if does what it was originally intended to do, can transform education in India and with it the quality of our workforce and
ability to compete.

What tech is supposed to do:  The internet and the tech startups around it, were supposed to transform business by doing the following:

-Reduce costs by replicating data or infrastructure (e.g. textbooks or classrooms) digitally, thereby  improving quality (i.e. high quality e-lectures replace low quality teachers in rural schools).
- Making information more convenient to access (e.g. learn at your pace).
- Allowing consumers to access to the `long tail’ – products that might be too expensive to distribute in physical form due to lack of demand e.g. books on kindle, or lectures on a subject with little demand, delivered online.
These advantages increase demand, by lowering product costs.

In reality, Ed-tech (like other consumer related startups) are available to only a small proportion of potential customers, who are already accessing the same product in another form (offline). Instead of democratizing education (which is the aim of all free state run schools), by making quality education available to all, at low cost, it stacks the deck further in favor of those students with higher incomes. Students who can afford to pay for both the physical infrastructure of a good school (compared to free state run schools) AND pay for learning outside of school, either through online or offline learning.

Ed tech in India (as per an industry report of 2023, that might over-estimate the opportunity, because its subscribers include investors) is estimated to have the following paid user base:
6.7 million students in K12.
2.8 million in test preparation.
7.2 million adults (between skill development & online courses and may include test preparation).

By comparison, 265 million students in India attend school.  68 million of them are enrolled in secondary or higher secondary schools. Only 10% of these (6.7 million) are the market for paid ed-tech products designed for K12 students (assuming people pay for ed tech products in secondary school).

The total number of households in India estimated to be middle class and above is around 23 million households. As a proxy for e-commerce products, Zomato has 16.6 million active users (if 1.5 users per household is a proxy, that gives 11 million households, of which only 1.5 million households are high value ). There are also just 23 million income tax payers in India ( or 15 million households assuming 1.5 tax payers per household) which is another proxy for market size. If every third household has a child in secondary school, that would be a market size of 8 million households, of which 6.7 million already consumer ed tech products. Byju’s, with no constraint on marketing spend, enrolled 150,000 paid users a month, or 7 million cumulatively – including repeat customers.  These are students already attending higher quality (paid) schools.

Far from being a billion people market, the (K-12) market is close to saturation. Any further growth comes with a higher customer acquisition cost and lower margins – at a time when the industry is cumulatively loss making. Byju’s losses have increased with an increase in turnover, even for the financial year ending Mar 23, when they had enough time to figure out their business model.  

This is a strong moral argument against ed-tech products in their current form, because it increases the role of money as a factor in a student’s success, rather than solve the problems tech is designed to solve,
and increase access for more people by lowering cost.
That was a consideration why China imposed restrictions on education and text prep products in 2021. An argument in China, which would be relevant in India, is that if everyone has access to the same ed-tech product (e.g. test prep) there is no improvement in outcome for a subscriber of the product.

In India, those with an advantage by virtue of being able to pay for higher quality education, would not have an advantage from ed-tech products withing their group, though it would increase the gap with underprivileged students (instead of reducing it). Restrictions  in China (resulting in a 90% fall in valuations) included not making a profit and restricting the number of hours of total learning. 

The (ignored) elephant in the room:

The largest customer for most services in India is the government, it’s also probably the most neglected
by startups. The largest e-commerce site in India, is the Govt’s GeM. (Govt e marketplace), In the current financial year (Apr 23 -Feb 24) the GeM has a sales volume of 317000 crore. This is almost 10 times the size of the entire Indian ed-tech sector in 2022. The GeM had a turnover of 201000 cr in the last fiscal year, so even at this huge volume, it grew at over 50% year on year. This is roughly the same GMV (gross merchandise value) of Flipkart or Amazon in India – across their businesses.

Similarly, the govt’s IRCTC website for rail bookings has a turnover of Rs 3540 cr, with a profit after tax of 1006 crore. The biggest non govt travel portal in India, - Make my trip, on a slightly higher turnover $ 593 million in f.y 23, had a loss of $ 11.2 million – both are listed companies, with MMT funded by reputed international investors.

In cities where the State run ONDC operates, the run rate of orders for ride hailing and food delivery
in the first year of operations, is already at the same level as the market leaders reached in 5 years.

The point of the comparison is that it is difficult for the private sector to compete when competition is free (or low margin). The private sector underestimate the ability of the govt to run a business efficiently and profitably. Rather than wait for margin erosion that competition from the government will invariably bring (just when several startup categories are just getting into profitability after years of losses), startups can look at the opportunity that Govt initiatives can offer.
 
In Education, the Central govt spent 112000 crore in f.y 23. The State governments collectively spent more. My state, Karnataka spent 32000 cr. in f.y 23. Two states, Maharashtra and Karnataka, together spent as much as the Central government on education (112,000 crore). By contrast, the revenue of the Byju’s (the biggest Indian education startup), across its businesses, including outside India, was 6500 crore in f.y 23. 

The real problem to be solved.
There are 2 reports that assess school education in India.
The NAS (National achievement survey) which covered 3.4 million students in 118000 schools in classes 3,5,8 and 10). &
The Annual status of education report (ASER). This covers the rural 5-16 (school going) age group, with the 14-18 rural age group covered in alternate years, irrespective of education level, to assess functional ability in language and math.
 
The NAS survey appears to have been `gamed’ by some of the states and averages are deceptive. However, it offers a large sample size and a large number of data points that validate each other.
Some trends stand out across states:

According to NAS 2021, learning outcomes were declined as students reached higher grades. Thus in mathematics, an average test score of 306 (out of 500) was recorded for class 3. This fell to 284 in class 5, 255 in class 8 and 220 in class 10. Tests in English, the local language and social sciences followed the same pattern. Our learning outcomes get worse once we get past primary education.

Across states, between 20 & 70% of students do not meet the minimum requirements for class 10.
Only between 8 & 50% met the required standards across states.
Specifically for Math (since an average across subjects is misleading) in class 10 in Delhi – arguably the state with the best facilities and where the State govt has made the improvement of state run schools a priority, 24% of students do not meet a minimum standard. However, the difference was 30% (clubbing the top 2 or bottom 2 achievement levels) between Delhi’s private schools and govt schools. Thus while 16% of students meet or exceed proficiency levels in Delhi’s state run schools, it’s 46% in private schools. 
In Punjab, which has the best rating in the country, with well documented state interventions, 14% of students do not meet a basic level of knowledge in Math (compared to 24% in Delhi). In UP, with below average scores, 34% of class 10 students do not meet a minimum requirement for Math.

25% of students across India study in a language different from their mother tongue (or spoken at home), which is a barrier to learning. In states with local languages that have very little utility outside the state, a majority of students in some states do not meet the minimum requirement (at least 35% correct answers) in a modern Indian language like Hindi. Thus in Kashmir, where tourism is a major source of revenue and where Kashmiri is not spoken outside the state, 67% of students in class 10 do not meet the minimum standard in any other language. (77% in Sikkim).

Schools run by the Central govt, had scores 10-15% higher than those run by the state govt, possibly because of better infrastructure provided by Central run schools (which are typically in Urban areas) compared to state govt run schools mostly in rural areas.  Urban schools on average did 5-10% better than rural schools. ICSE and CBSE schools (with better infrastructure) did 20% better than state board schools.

Half the teachers across states have not had on the job training and only 25% have adequate teaching material. While less than a third of schools have internet or computer access, 90% of Class 10 students have access to a smartphone, which is an opportunity.

Though ASER figures look different, they are consistent with NAS findings for class 10. Around half of all the 14-18 year old’s surveyed were functionally illiterate – unable to read, or do arithmetic at the level of a class 2 or class 3 student. 20% of those surveyed, had either not attended school, or dropped out and since a third of those that did study in class 10 did not meet minimum standards in NAS tests, the NAS data is consistent with ASER.

Thus the problem in Indian education is that half of all students at Class 10 level are functionally illiterate, due to a combination of poor school infrastructure, poor quality of teachers, or lack of understanding of the medium of instruction. While tech can bridge the gap in all these areas, Ed-tech is focused on the 10% do exceed the required standards (because they can afford better quality education). The problem is amplified when appearing for employment related tests. Most of those that apply will have no chance of meeting standards and the 10% that do are paying more to get a competitive advantage (through ed-tech subscriptions) over their 10% peer group, who are also buying the same services – which is why everyone who has cleared the JEE or UPSC appears to have taken 1 or more coaching classes.

Swayam: As mentioned earlier, the govt is the biggest user of tech solutions. The Govt’s ed tech portal Swayam is little known (none of those I did an impromptu survey with were aware) but has more reach than the heavily funded Ed tech companies. The Swayam website shows that it has partnered with leading education providers, including the IITs and IIMs. 

The Jan and July semesters of Swayam have over 3.5 million registrations. The Jan 24 semester has 3.7 million registrations.  15-20% of those who register, take the exams (for which one has to pay) and 10% of students have a paid certificate – compared to barely 5% for MOOC courses internationally. The number of registrations annually for Swayam (even considering duplicate registrations for the Jan and July semester, is more than what the Ed tech market believes is the market size for paid courses for adults – with a 0 customer acquisition cost for Swayam. The Swayam (Plus) portal is extending its offerings, in partnership with leading companies like Microsoft, google and TCS, at a much lower cost than what has hitherto been offered by Ed tech startups at a relatively high fee. A logical extension of Swayam would be for e.g. to offer free coaching for entrance exams - something individual states could do (in the local language) or which could be offer to test takers from lower income families. This is something that could badly affect existing startups in the test preparation space.  

Although Swayam has partnered with NIOS and NCERT for school level courses, there are none for students upto class 10, which is served by the other Govt initiative – Diksha, though `Swayam Plus’ is actively looking to expand offerings.

Diksha: Diksha is intended to be the single portal for school children and teachers, which in theory covers 173 million students (though user details are opaque). Diksha mostly comprises 6477 textbooks in digital form, across boards and languages and training resources for teachers. It is free and developed by the EkStep foundation and has open source software, which would enable State govt’s to build their own offering on it – through the open source platform Sunbird.  The Diksha platform has been introduced in a dozen African countries.

Assuming a user spends 100 hours per year on the app, or 2 sessions per week, there are probably 7 million Diksha users across India. The usage varies across states. Gujarat has 100,000 user sessions a week, whereas neighboring Rajasthan, with a similar population, but less developed, has 3 million – and coincidentally has risen in NAS rankings.  In the south, while Karnataka has 1 million sessions a week. Andhra has 3 million, while neighboring Telangana and TN have only 100,000 sessions per week. UP has 3 million sessions per week, while neighboring Bihar has 200,000.

The variability in usage is attributed to State govts not having the wherewithal to develop applications on the Diksha platform, which presents opportunities for startups. 

The opportunity:
A Ed tech startup can produce local language content (same lessons as their syllabus) and partner with a State govt (Diksha provides a platform), to introduce it to students free of charge, through their cell phones, with WhatsApp groups for peer learning.

A pre-recorded lecture of a class, with a high quality, highly peer rated teacher, teaching in a language the student is most comfortable with, will resolve the problem of poor quality of teachers and poor class infrastructure. ASER surveys show that 40% of students are paying in some form, for extra tuition. This willingness to pay for quality education can be used in a premium version, with (for e.g.) a WhatsApp group for group tuition, administered and taught by a local teacher with a high student rating.     

 A subscription of Rs 100 /month for the WhatsApp group (with 100 participants) will compensate a teacher for 2 hours of time each day. There should also be a subsidy from the state govt which will compensate the ed tech company – apart from being paid for their content in recorded videos. There can also be an incentive based bonus, based on board exam results, or passing a state govt standardized test (which the ed tech company can administer).     

There are specific areas where a State govt would like to hire additional teachers on this basis (a limited subsidy which will be a fraction of a full time teacher) supplemented by subscription fees. For e.g. in the state of Jammu & Kashmir, most teachers were traditionally Kashmiri Pandits, who fled the state. A lot of schools in rural areas were burnt down, but students repeatedly got promoted, leading to a situation that the ASER report describes in other states – where an 18 year old, who has passed class 10, is still functionally illiterate. There is a need to rehabilitate Pandits (for which the Central or State govt will be happy to provide a subsidy) yet they cannot practically return to the state. Working as part time teachers, teaching in Kashmiri will help some of them and improve the skills of thousands of Kashmiris and make them more employable in an environment where normalization of the law and order situation will create thousands of jobs in tourism. 

There is a similar opportunity in the North East (with the lowest NAS scores) or in remote hill districts,
where students are not taught in the language they are comfortable with. In Nagaland, the medium of instruction is mostly English. While NAS scores for English are slightly higher than the average for India, the scores for every other subject are very low – because students and teachers are comfortable in Nagamese (itself an amalgamation of tribal languages) but have to teach and be taught in English. Thus only 6% of Naga 10th Class students meet the required standard in Math and 3% in Science. It is a similar situation in Arunachal (with English scores lower than Nagaland). Across India, the problem with many English medium schools is that the quality of English when other subjects are taught, is low, which, when coupled with the student’s lack of exposure to English, results in low NAS scores.
Goa is a case in point, (low NAS scores in a State with above average school infrastructure) where Church run English medium schools are under some pressure to teach in Konkani. 25% of students can potentially gain from an app which teaches them in their mother tongue. Another 25% who study in the schools with the worst infrastructure (excluding those in the `different’ language category who also fall into this one) can immediately gain from a tech solution.

The low quality of teachers (and the lack of resources to train them) is another potential opportunity.
The NISHTHA section of the Diksha portal has teacher training resources but they have shortcomings. Teachers could be made to undergo self-paced sessions which can cover both teaching techniques and the lessons to be taught, with clearing promotion being contingent on clearing an online test. The CBSE board mandates that a teacher spend 50 hours on training. Online training on a separate CBSE portal has been seen to deliver that at a fraction of the cost the govt is prepared to pay for offline training.

There is a potentially large opportunity in what I would call a `back to school’ segment. Many of those in the 14-18 age group who are functionally illiterate, have parents who are either illiterate, or not educated enough to understand a child’s lessons. About 25% of parents of school going children in India are illiterate, as per NAS. They might want to be educated but cannot attend school – due to domestic chores, or, even if the facility was available, a mental barrier to studying alongside children.

It may be possible, with the involvement of the state govt, to have the equivalent of a 10^th standard diploma in specific subjects that can be taught online with exams (as is the case with the Swayam portal). A subsidy for this can be drawn from into Govt schemes to for e.g. empower women, or raise literacy levels. Such a scheme lends itself to offering relevant subjects not from a child’s syllabus e.g. agricultural practices, or health and wellness. Taking a number of these courses can qualify as passing Std 10 (or 12), with an option of diplomas in individual subjects. 

There are Ed tech startups that offer products for `Bharat’ along these lines.
Vidhyakul offers lessons for Class 9-12, in 3 states (UP, Bihar and Gujarat) with subscriptions starting from Rs 350 per month, with a free access option. They have separate classes for spoken English and IIT entrance. They have shown that you can be operationally profitable at a subscription of Rs 350/ month and a customer acquisition cost of Rs 250 and just 1 cr. revenue per month. They have demonstrated their success – by the number of district toppers in UP and Bihar (who make better brand ambassadors that Lionel Messi). More importantly. Vidyakul has accessed a market untouched by existing better funded players. 95% of students using Vidyakul, have free access. While there is an untapped segment at Rs 350/month, the bulk of the market cannot pay and will need the State to do so. 

Another ACT funded startup is VOPA, (funded by a philanthropy platform ACT, which funds similar ventures to transform education) which offers free online learning in Marathi by partnering with local communities in 7 districts with 1.7 million students and teachers enrolled.

One of the ACT funded startups, English Quest, has a 70% retention rate with schools who pay for it and 85% of students achieved the desired level of English knowledge. These are adequate grounds for a State govt to fund such a program.

ACT’s and EkStep’s thesis is that for Ed tech to have a meaningful impact in India, it has to be low cost, in vernacular and mobile related – Do any of the heavily funded (and loss making) Ed tech startups in India meet these criteria ?