The hard thing about edtech products
A product is something you build with a user in mind. Users are the difference between a working software and a product. Therefore, all products are measured by how well they serve some group of people, and how many people use them. There are, of course, products that may not do a great job of serving their target user and are still heavily used by large numbers of people, because the users have no other options. To me, the education system is a good example of that.
We attend school from the time we are 5 or 6 until we’re in our 20s or 30s. If you consider the education system a product, we’re the DAUs (daily active users) of that system for about 20 years. We are the consumers of that system, that product, for 20+ years. Fantastic adoption and retention by all means, though with a debatable efficacy and efficiency.
Efficacy and efficiency
Some people say it’s okay that Computer Science faculties globally unintentionally end up focusing on the top 20% of their students, leaving the middle 60% on their own, and justifying the failure of the majority by referring to the top 20%’s success, which could potentially explain why Computer Science drop out rates in Germany is 50%. I think that’s far from okay. If anything, it’s the education system’s failure to learn from its input and output year on year and develop a playbook that improves the quality of education for each new incoming generation of students.
Retention and adoption
Regardless of what we think about the education system, I’m gonna reiterate that it does a great job of retention. The education system is essentially a universal system of well aligned extrinsic motivations. Education’s customer churn rates never get bad enough for the system to self-improve, because the cost of quitting is too high, and so is the network effect. 20 years of schooling gives us a certificate, which gives us a job, which gives us money, which gives us a life. “Are you sure you wanna do this to your life?” is the question people will ask those who consider dropping out and not “Are you sure you wanna stop using this system?”.
There are great reasons for going to school, whether or not it works. The “why” is damn clear.
Building an edtech product in 2021
Many edtech companies over the past decade have assumed that people will use their product because people are already using schools as a product. That’s not automatically true.
If we’re gonna build an edtech product in 2021, we still need to answer “Why would people use this product?”
The question of why is a question of motivations. Motivations can be extrinsic or intrinsic. Schools and universities have a monopoly in delivering on extrinsic motivations.
Intrinsic motivations, however, are still pretty untapped. I’m not referring to the shallow gamification tactics like that of Duolingo or dark UX and advertising measures that most tuition startups in India and elsewhere employ to sell via fear or fomo. I’m referring to healthy measures that we humans are wired for. The things we do generation after generation that we’re not becoming better at, that we’re not receiving any sort of training for, and that we can’t simply ignore or quit.
What are we intrinsically driven to do?
Things that activate our curiosity and/or needs. For instance, parenting is a journey we sign up for that we’re not trained for, we’re not good at, and we can’t quit. We’re intrinsically driven to wanna do well at it. That’s also the case for starting a new career without a playbook, which is basically most high growth careers in 2021.
There are some other sort of behaviours that we’re probably biologically wired for, such as nature connection.
How can technology enhance things like parenting, nature connection, career growth and etc.?
The first step is really problem definition. We need to acknowledge that there are problems worth solving in these domains, and create the means for solving them.
If parents of the world don’t have enough time for both themselves AND their kids and they’re hiring iPads and smartphones to temporarily get rid of their kids every day so they (the parents) can do what they need/want to do, why aren’t there deep tech companies out there that’d multiply parents’ ability to grow themselves and their children?
If people are clueless for the first 3–4 years of their careers, why don’t we have a deep tech, fast learning category of companies called “Early Career Education” that can make universities obsolete?
If we’re spending 20 years at school and less than 2 months in the nature and absolutely ruining the environment as a consequence of that, while innovating new types of mental health disorders as a generation, why aren’t we making deep tech companies that take learning and education back to the nature?
What does it take to disrupt education?
I’m gonna share what I think it takes to disrupt education but our actual ability to execute on that as a civilisation depends on whether some of us are willing to dedicate the next 30+ years of our life to building the next education system.
On a side note, if you know anyone who’s that committed to creating a bottom-up change in education, please connect us.
To disrupt education from the bottom up, patiently and profitably, it’s essential to find a specific segment within the current education system’s 20-years-long user journey, make it 10x better, and use that base and profit and go capture more of that user journey by more 10x improvements at each stage.
Once we identify a specific segment in the broader education journey, it’s essential to win the users’ attention twice:
- Once for product/market fit
- and a second time because learning itself is an attention intensive task
An example of an edtech product that has decent product/market fit but doesn’t teach you anything: Duolingo
An example of an edtech product that has great content and teachers but terrible completion rates and subscription renewals (to the point that they had to partner with universities and give away certs for survival): Coursera.
Winning users’ attention twice
The key isn’t to try to win users’ attention twice, but find a problem space and/or a solution that makes one of these 2 types of attention a constant, so we will have 1 constant and 1 variable, instead of 2 variables. Coursera and Duolingo tried figuring out 2 variables at once, and they failed at that, so now they’re trying to make one constant, but at this point they’ve built a huge organisation around that old way of operating, with a goal setting and culture and engineering and product infrastructure optimised for that old game. Hence their inability to innovate (I do hope they can find a way out and innovate again, though, it just sounds really unlikely unless they have some really great leadership).
Product/market fit
This follows the same playbook as all other products: a product that a large enough group of people will consistently use that makes them better that makes the creator $X in value where the creator keeps Y% of X, with X and Y being independent variables. It’s not easy, but it’s simple.
A learning experience that works
This is the science of pedagogy and the place where we lack innovation. Pedagogy is essentially an equation for learning, except most people who’ve studied pedagogy academically didn’t study much math and/or stats and/or systems thinking at college; so they’re not naturally able/willing to look at their job as optimisation. They’re often too stuck on the models they know and aren’t able to communicate learning experience in terms of constants and variables that we can tweak.
Innovating in pedagogy is a must, and to get this job done, we either need to train/hire people who are hyper versatile (understand pedagogy deeply, are systems thinker and understand UX and adoption), or we should build companies that are inherently great at facilitation, starting from the leadership itself.
It’s about facilitating a conversation between product people, pedagogy experts, engineers, users (and potentially investors) to create a product that works, makes money, and enables continued disruption of the rest of education.
Continued disruption of the rest of education
The ultimate goal, to me, is personalisation of education, so that instead of one education system for everyone, we can have one education system per person, so that each person can have the means to be the best they can possibly be, given that they choose to do so.
I have a TEDx talk which I recommend you watch if the above paragraph made you think of Star Trek (I haven’t watched it but I’ve been asked a few times if that’s how I mean personalisation, and based on the explanations I’ve heard, the answer is no).
Personalising education
We’re lucky to live at a time when there’s a playbook for personalisation, and companies who do it for music and TV show recommendation.
At the highest level of abstraction, in order to build the world’s next education system, one that’s personalised, we need to kickstart the whole AI product value chain sooner than later. That is, a recurring behaviour, a product that quantifies the behaviour and makes it available and accessible, data, prediction models, prediction-centric value generation and capturing at a niche scale, scaling up the data diversity and prediction context/relevance in order to or as a consequence of scaling up the category in which the business operates. For instance, start niche in parenting or nature education or whatever else, collect some data, deliver some prediction-centric product that sells and grows the customer, then use that to learn more about the customer, broaden the context of personalisation and data collection, leverage that to capture new types of customers, repeat and so on.
What sort of model?
There are probably more than one way, but one of better ways I’ve come across is presented in this paper. To make that model work, the following data points are needed, at some scale:
The data
- data about some learning goals
- data about the prerequisites of each learning goal
- data about learners’ competence with respect to the prerequisites of a a learning goal they’re into (or one that we think they may be into)
- data about what different learners have learned
The faster we acquire this sort of data broadly, and the faster we develop new layers of depth in each of them, the sooner the disruption of global education will be viable.
To recap
To build the world’s next education, we need a mixture of pedagogy, product design and development, and AI value generation and capturing, and probably founders who are crazy enough to commit to this multi-decade journey.
That’s the learning journey that I’ve signed up for daily, every single day for the past 5 years, on my happy days and also through the extreme ups and downs of my life. Last week I moved to Singapore on a tight personal financial runway to join Entrepreneur First Singapore to start a company and be a first time CEO (I still struggle to answer the question of “What role do you wanna have in the company that you’re building?” because frankly, I don’t care. I care quite exclusively about the outcomes that I described in this post, and the only reason I wanna build a company is because I don’t see any existing edtech companies worth their salt.)
Equally scared and excited as my journey gets more real than any time before from here on. If you’re a coach, pedagogy nerd, or someone who just wants to be part my journey and enable me and grow with me, please reach out: http://linkedin.com/in/sinameraji
