Cost vs Quality in Edtech - Keith Schacht, Avichal Garg, and Geoff Ralston

by Y Combinator4/6/2018

Keith Schacht is the cofounder of Mystery Science, which makes lessons that inspire kids to love science. They were part of the Summer 2017 YC batch.

Avichal Garg is an Expert at YC and prior to that he was the Director of Product Management at Facebook.

Geoff Ralston is a Partner at YC and before that he cofounded Imagine K12, an edtech accelerator that’s now makes up YC’s edtech vertical.


Google Play


Craig Cannon [00:00] – Hey, how’s it going? This is Craig Cannon, and you’re listening to Y Combinator’s podcast. Today’s episode is about EdTech and is with Keith Schacht, Avichal Garg, and Geoff Ralston. Keith is the co-founder of Mystery Science. They make lessons that inspire kids to love science, and they were part of the summer 2017 batch. Avichal is an expert at YC, and prior to that he was the director of product management at Facebook. Geoff’s a partner here at YC. Before that, he co-founded ImagineK12, an EdTech accelerator that now makes up YC’s EdTech vertical. All right, here we go.

Geoff Ralston [00:32] – Avichal, you founded PrepMe in 2001, sold it 10 years later in 2001. That was actually the year we founded ImagineK12, the world’s first educational technology accelerator. Keith, you founded Mystery Science, I think in 2013.

Keith Schacht [00:51] – We just celebrated our fourth birthday in January.

Geoff Ralston [00:53] – You just celebrated your fourth birthday with Doug Peltz. You guys actually did YC in summer 2017. So the idea of this podcast was to have a conversation about educational technology, about maybe some current, past, and future stories around EdTech. One of the things that started, that was the initiative for this conversation was a blog post that Avichal wrote in 2011 titled Why Educational Startups Do Not Succeed. His summary of that post, I’ll just give it really quickly since he helpfully put it in the post itself, was that entrepreneurs in education built the wrong kind of business. They think of it as a quality problem when actually it’s a cost problem. Building an education does not follow an internet company’s growth curve. Because of that, you have to think way longer term than venture capitalists are interested in funding you. The opportunities for servicing education are actually not in the US. They’re actually in Asia. It’s not really going to be relevant for another five years. That was in 2011, so maybe it’s relevant now, cause that would have led us to 2016. Maybe, Avichal, you can start by sort of reflecting on why you wrote that post and how you’re thinking about it today. Hopefully you guys can start having a conversation about that.

Avichal Garg [02:26] – That’s a great summary. It’s a great intro. I mean, the genesis of that blog post, for us, was really we started this company. We had a lot of ideas around how to make education better using machine learning and using technology and personalization. It turned out a lot of people wanted that in the early days, and so we got some early signal that that’s actually what the market wanted. Then very quickly things started to plateau, from a growth perspective. My conclusion was that we’d saturated a lot of the early adopters. The mainstream market just wanted different things. The mainstream market really cared much more about cost rather than quality. The fundamental sort of disconnect between people who start education businesses and the consumer themselves is fundamentally that. The people who start education businesses tend to care much, much more about the quality of the product. In the US, the average consumer just looks at it differently. The buying psychology is fundamentally different. There are a whole host of reasons why that’s the case that we can get into, that I think actually have very interesting downstream ramifications on things like politics.

Geoff Ralston [03:32] – About whether people value education-

Avichal Garg [03:34] – Yeah.

Geoff Ralston [03:35] – In itself.

Avichal Garg [03:36] – Exactly. If we want to sort of unpack that, the thing I think, and I have very little other than anecdotal to back this up, so this is my current belief. I could be wrong. I’d love to hear, probably from Keith, why I’m wrong. But I think what happened was post-World War II America was really good for a lot of people. I grew up in Ohio and Kentucky. If you’re in the 60s or the 70s, you could get a job at the GM factory, you could make a good wage, your spouse could make a good wage. Maybe together you make $80,000 a year working the factory, and a house cost $80,000 a year. That would be like living in San Francisco, and you and your spouse make like three million dollars a year, and you never went to college. I think rationally you would look at that and be “Why would I go to college? I’m making three million dollars a year. I can buy my house, and I can have a great life.” Then the people that were sort of born, just to make the math easy, let’s say you were born in 1982. You’d be 18 in the year 2000. You’d be graduating from high school. The people who were born in 1982 were raised by people of that generation. The value system scame from a belief that the world would still work that way. But the world doesn’t actually work that way, right? Starting with 2000, the internet, with the crash in 2001, the massive, there was, in between, little recessions. There was a crash in 2008. Basically if you were an adult in 2000, you’ve kind of been in and out of work

Avichal Garg [05:03] – for most of your adult life, if you were kind of in that camp, if you didn’t go to college. That has bleed over effects into things like politics, where the system, the way you thought the world was going to work, the world doesn’t actually work that way anymore. That disconnect has been a really hard transition. There’s this entire sort of 15 to 20 year range of adults that kind of got trapped by it. I think we’re seeing some of the effects of that right now in politics.

Geoff Ralston [05:30] – Keith, maybe you can react to that. I’m especially interested in this perspective, which I think was more true in 2001 than it is in 2018, which is that if you’re going to make any money in educational technology, you better appeal directly to parents, because schools weren’t going to pay for technology. Actually, they’d been burned with a whole bunch of hardware, and technology wasn’t so relevant to them at the time. That seems like it’s shifted a bit. So maybe you can-

Keith Schacht [05:59] – One interesting thing when talking about the space is, it’s actually hard to talk about the education space. Elementary versus high school are totally different markets in many respects, and parents versus institutions are very different markets. It’s interesting. We often sort of group this as the education space or the education industry. We have to be careful about generalizing. Just one line, in Mystery Science, we are helping fix science education in this country by giving elementary teachers a remote science expert to co-teach the class with them. We effectively sell schools a curriculum that replaces the textbook. You can think of it in a sense as a digital textbook. Our whole pitch to teachers and to schools is that this is better. This is better education. It’s better than the textbooks, it’s better than whatever the conventional curriculum is. It’s notable that our teachers and the students and even parents, increasingly, are starting to recognize that it’s better. It does give me pause to hear, you can’t build a business by making something better. I think there are some interesting challenges to trying to, how you communicate that your product is better and how the customers appreciate it. For just very early on, we’ve gotten a lot of traction selling it to schools and getting appreciated for making a better educational product.

Geoff Ralston [07:24] – It does seem to be true, to back up Avichal’s point about parents, in general, not being willing to pay, especially sort of the mass market, to be willing to pay for an education quality. The value that they place on that, in the United States, at least, is still not clear. There’s a lot of money being spent outside of school on education. But it’s test prep and tutoring for pretty affluent families. Isn’t that true?

Keith Schacht [07:57] – It is an interesting, so we are preparing for a transition from selling to schools, selling direct to parents. We’ve sort of dipped our toe in the water with homeschoolers, which is a relatively small market, and are getting ready to go mainstream. There’s still a lot yet for us to prove about that. But there’s a couple interesting observations that we’ve made so far. The thing that makes the market a funny market, the parent education market, is that you get this product for free. Basically through property taxes, that’s how you pay for school. But as an entrepreneur trying to build a business in the space, it’s interesting to compete with free out there. Cause parents feel like, “Well, I have this thing.” One thing we’ve noticed in interviewing parents is that at the very young age, parents feel very involved in learning. From birth to when they send their kid off to kindergarten or preschool, the educational value of the products they’re buying is one of the dominant factors in considering whether you’re going to buy that speak and sound or that play mat or that book or whatnot. It’s a huge factor in buying decisions. Once you send your kid off to school full time, parents remain involved in some respects. Reading is one of the main ones. They still feel like it’s their job to make sure they’re reading regularly to their child and helping to reinforce their ability to read at home. Then it sort of drops off for the next like six to eight years. Your involvement as a parent in education

Keith Schacht [09:26] – is making sure you’re supporting the school and going to parent-teacher conference and picking the right neighborhood. You sort of get involved in that respect, not by buying products and sort of delivering education at home. Then towards the end of high school, it kind of ramps up again with the whole getting into college thing or shoring up a weak grade or whatnot.

Geoff Ralston [09:44] – Well, it certainly seems that your prediction that venture capitalists wouldn’t fund EdTech companies has not come true, especially since 2011. It’s very interesting. We founded ImagineK12 in 2011. The companies that we funded alone received hundreds of millions of dollars in venture investment. Obviously venture capitalists saw opportunity somewhere, right?

Avichal Garg [10:12] – They basically made the mistake that I tried to call out in that post, which was they expected that the ramp of those companies from a business perspective would look like an internet company ramp. I don’t know the numbers of most of the American companies, but they pale in comparison to Chinese education companies that go from zero to a billion in IPO in seven years. It’s just like the ramps are just crazy in a place like China.

Geoff Ralston [10:43] – A lot more people in China, too, as it turns out.

Avichal Garg [10:46] – That’s part of it, right?

Geoff Ralston [10:47] – The public school system is much worse.

Avichal Garg [10:49] – Exactly. The cultural dynamics are different. The public school dynamics are different. There are a lot of factors there, for sure. But the net effect of it then ends up being on paper you might have a company here, as an education company, that’s worth 100 million dollars or 200 million dollars and has raised a lot of venture capital. But do the revenue numbers actually back that up? Could you actually take that public on the market in two to three years at a billion dollar valuation, and what kind of revenue would you have to do to support that? By and large that hasn’t been the case, which is why you haven’t really seen education company IPOs.

Keith Schacht [11:18] – But is that unique to education companies? Because just take consumer tech generally. There’s many hundred million dollar plus valuation consumer tech companies that don’t have the revenue to back it up. There are also many consumer tech companies in China, where the Chinese equivalent will scale much faster than the American equivalent.

Avichal Garg [11:35] – It’s certainly true that I think you have a lot of companies that end up in this kind of dead zone. But the reason they do is because there are some that make it out. Every now and then, you’ll get a Snapchat or a WhatsApp or an Uber or a Lyft or a Stripe, or the list goes on. Most of these companies are not that old, right? These are all seven to eight year old companies, right? So it’s the hope that you stumble into an Uber or a Stripe that really, supports the rest of the ecosystem. But I don’t see, where’s the Uber or the Stripe of the education space?

Keith Schacht [12:07] – The closest is probably ABCmouse, recent billion dollar valuation, straight to parents. Teachers Pay Teachers would be another one.

Geoff Ralston [12:19] – Chegg is worth two and a half million now.

Keith Schacht [12:21] – Chegg.

Avichal Garg [12:22] – That’s actually a great example. That’s actually the one that I use most often, which is, by and large, if you look at Chegg, the way they got to where they are is they didn’t reinvent digital learning. They just made textbooks cheaper.

Geoff Ralston [12:30] – They’ve almost completely shifted. If you look at Chegg’s website today, they don’t talk about textbooks anymore. It’s they’re a smarter way to student, whatever that means. It’s all about digital education for Chegg now. I know it’s a multi-hundred million dollar business for them.

Avichal Garg [12:51] – I was going to ask, do you know how that number breaks down between their core textbook business versus the digital-

Geoff Ralston [12:56] – I think the core textbook business and volume is bigger, but the digital business is-

Avichal Garg [13:00] – Growing.

Geoff Ralston [13:01] – Rapidly growing and scaling. So it’s interesting. There are EdTech companies that are scaling. Certainly, revenue has been tough for many of them, although, I look at companies like Mystery Science, which has actually scaled revenue, without mentioning any numbers, very well by selling to schools. I do think selling to parents still, because of this free zone, and we have to be clear that for better or worse, that free zone is for the mass market middle America. There is plenty of more affluent families that spend tens of thousands of dollars every year on private schools, on private tutors, on SAT prep, on all sorts of educational advantages that do constitute a pretty big market, right?

Keith Schacht [13:56] – It’s notable that ABCmouse is, in some ways, I think of them as the new Leapfrog. They’re targeting that same age demographic that Leapfrog used to, which is outside of that free zone.

Avichal Garg [14:07] – It’s the early stage. That pattern has held for a while, too. That’s why in that blog post I called out something like five years, because what you’re starting to see is, cultural shifts just take a while, right? There is a cultural shift. The cultural shift, essentially, is that generation of parents that were 18 in the year 2000, they’re parents now, and they have five year olds. The person that was 18 in the year 2000 would be 36 today, so they might have some young kids. They’re starting to look back at, and they probably don’t have a 25 year old, but they might have a six year old or an eight year old. Starting to look back at their life and saying, “Wait a second, what happened? Oh, the world changed.” People are generally pretty smart. That cultural change is starting to happen. That might be the 36 year old that’s now head of the science department and is looking at it and saying, “Wait a second, I grew up with the internet.

Avichal Garg [14:58] – I get this, we should be buying better stuff.” That change is happening, but we’re kind of at like the early parts of that change. Hopefully it accelerates. But I don’t think the mainstream has flipped quite yet.

Geoff Ralston [15:10] – There’s another aspect to the cultural change, which seems to me is a cultural change within schools, which for me has shifted radically. In 2011, when you wrote that post, the fact is that most schools were not connected with reasonable internet access. Today, almost all of them in the United States are connected with enough bandwidth for more than one person could run YouTube at the same time. Many schools have directors of technology and are thinking about their technology buying plans. I don’t know how you think about this-

Keith Schacht [15:46] – In the school space, that’s one of the most significant things, is from my perspective, it seems like it is mirroring the enterprise cycle, lagging about 10 years or so. When we started four years ago, we finally could assume that most of our teachers would have a computer, one computer in the classroom, a projector so the kids could see it, and an internet connection. Not the best internet connection, often consumer-grade, shared. It would slow down during lunch hour and then speed up afterwards. We could assume this baseline level. That was 2014. So maybe that only happened around 2012 or so. That’s relatively late compared to what happened in business. We are really deep into this first generation cycle, where there’s a whole wave of companies that are just bringing efficiency to schools. Now that we’re on the internet, we can make attendance easier. We can make grading easier. We can make student rostering easier. We can make a lot of back office stuff. That’s a whole wave of companies that aren’t even involved in learning. They’re just involved in making these business operations more efficient, cause, hey, software’s good for that.

Keith Schacht [16:56] – Then there’s the second parallel we’re seeing. This was part of the wave we’re riding, and other companies, where it used to be that there were directors of technology in business who made the software purchasing decision and then rolled out to everyone. That’s how the school director of technology started. Now we take the Salesforce sales model to these teachers, where Salesforce didn’t get permission from an IT department. They just got someone in marketing who signed up for a free account and got started, and then three or four or five more people got it. Eventually, the IT department had to take notice, cause they had 50 users in the company. We do the same thing with our software, and other EdTech companies, too. I don’t think we pioneered this.

Geoff Ralston [17:35] – The pull model instead of the push model. Instead of having to wine and dine supervisors around the country, you actually convince, persuade teachers that you have a great product. In the end, they persuade their administrators to buy.

Keith Schacht [17:50] – Yup, yup. There’s a bunch of interesting shifts that happen there. So Salesforce was selling against incumbent CRM providers. I assume Oracle had a product like that. Their product looks different. The onboarding for it looks different, and the cost structure looks different. We’re seeing that same shift. We’ve got the Pearson, McGraw, Houghton Mifflin, that generation of companies who still assume they’re going to sell top-down and roll out. Then Mystery Science and other EdTech companies that just do the a la carte and roll it up.

Geoff Ralston [18:18] – So Avichal, do you think your conclusions still stand today? Have you modified your take, given what’s happened over the past, you sort of gave it five years, and it’s been seven. Let me just read your conclusions. I’m wondering if you think they’re still valid today or whether you would council new EdTech entrepreneurs differently.

Keith Schacht [18:44] – This is the good and bad of putting things in writing.

Geoff Ralston [18:47] – Well, no, and to be fair, I disagree with very little of what he wrote in 2011. The real question is, have we moved on? If so, how should we move on? Your first point was don’t make a better product, make a cheaper product, which, by the way, this efficiency stuff that Keith was just talking about plays directly to that. Don’t start in the US or other developed countries, cause it’ll be too slow. Start in Asia. Don’t take VC money. Don’t target the middle class, because they don’t really care. Don’t expect a quick flip or quick growth. You want to, you can take those as a whole or one at a time, whatever you like.

Avichal Garg [19:34] – I can take them as a whole. I think they’re generally still true. But there are pockets where it’s starting to change. I think the places we’re starting to see change are especially around, I think schools have changed a lot. I was talking to Dan at Clever, and he was talking about how Chromebooks have just changed the landscape.

Geoff Ralston [19:52] – Dan, the CTO-

Avichal Garg [19:54] – Correct. Dan Carroll, the CTO of Clever. Also, is Clever a YC?

Geoff Ralston [19:57] – Yeah, Clever is a YC. Should be in ImagineK12. Long story for another time.

Avichal Garg [20:03] – So he was making the really great observation that, to your point, too, is that just changes the game. All of a sudden, you can actually assume that there’s an internet connection and a thing that looks like a laptop and is pretty reasonable and has good specs. The fundamentals have changed. But the way that the fundamentals then manifest in terms of startups is just starting to change, cause there’s a phase shift there. So now that internet is good, now that the person making the decisions might be the department head or the teacher, now that there are products that you can just sign up for, and you don’t need to go through the district, all those are structural changes that have just started to happen in the last, let’s say two years, three years, that it’s really gone mainstream. Looking forward now, the question really is, are they still true today, take a snapshot? The answer is mostly still true. But that’s not really the question that a startup should be asking or a founder should be asking, is will it be true in five years? I don’t think it will be true in five years. If you’re starting a company today, the question really is are there enough customers for you to get going? Has the world evolved such that in five years it will be dramatically different and look a lot more like you’re early adopters? I think it will.

Geoff Ralston [21:09] – Well, they’re all forward-looking statements. You’re basically saying ignore that blog post, because the future’s different. The future looks actually better for it.

Avichal Garg [21:21] – The future looks a lot better. The present, as of today, still kind of looks like that. But the trend lines are good. You look at Mystery Science or you look at NoRedInk or you look at Clever, all these companies that have managed to get pretty substantial traction. If you’re starting today, some of these markets are actually, you can get pretty reasonable growth pretty quickly. The part of it that’s still hard is actually the monetization. That part of it is just going to be phase shift. That will come. It is a case-by-case basis. In the early, early parts of the market, kind of what Keith was saying, if you’re targeting four year olds and five year olds and six year olds, you might actually be able to make a lot of progress and actually monetize, because the buying psychology has changed. If you’re talking about 14 year olds, you might not be able to. It depends on the market you’re going after. The Asia thing is 100% true. If you just look at the growth of companies and the revenue of those markets, China is just an amazing, amazing market. Unfortunately it’s not, most people are not faced with the decision of should I start a company in the US or should I start a company in China. It’s one or the other.

Geoff Ralston [22:31] – It’s difficult in US to go overseas first.

Avichal Garg [22:33] – And vice versa.

Geoff Ralston [22:35] – You just don’t know the market, and vice versa.

Avichal Garg [22:36] – That’s not really, for most people, a decision point. I think the more important decision point for most entrepreneurs probably is like which part of the market am I going after and why, and is it higher ed, is it high school? Is it science education, is it math education? Is it Pre-K? There’s a lot of ways to segment the market. Some of those have moved much more than others. The trajectory looks really good. There’re going to be pockets of pretty substantial change in the next five years. Some will just take a while longer. High schools will take a bit longer than some of these others, just cause the phase shift, right? The parents, you kind of have to have the market grow into it.

Geoff Ralston [23:11] – Keith, maybe we can talk a little bit about Avichal’s original, first point, which was the precise opposite of what you guys did later, which is don’t build a better product. Build a cheaper product. You guys, in fact, didn’t build a massively expensive product. But you built a better product. Your whole idea was to build better science education. This is an interesting conversation because the real issue in American education isn’t price. It’s quality. It’s how can we get better outcomes. Maybe you can reflect a little bit on that opportunity.

Keith Schacht [23:47] – There’s opportunity to build a better product in the education space. Right now, that’s absolutely true. It was true four years ago, even when we started, that you could build a better product. To the extent that the advice is useful, of don’t build a better product, build a cheaper one, it’s not unique to the education space. There are many, many startups whose growth is driven by lowering cost. The whole media space, the news media space online is being driven by free newspapers instead of print newspapers. Lots of people argue, oh my gosh, consumers don’t care about the quality of reporting anymore. All they care about is something that’s free. It’s true that free is a really important driver in every industry. Lowering cost is an important driver in every industry. That’s not unique to education. But the conclusion that the consumer doesn’t care about quality, it’s one I’m very hesitant to make. Because oftentimes we, as the creator, have a unique set of things that we think are important. Then we imbue that in a product. If it doesn’t resonate with people in the same way, oftentimes we have to charge a little bit more. Then it’s easy to think, “Oh, they just don’t care about quality.” I could image in the furniture space, just to pick a totally different industry, one of the high-end designer furniture stores for many years was Design Within Reach. Never actually shopped there, but I walked in one. They sell $20,000 couches.

Keith Schacht [25:16] – That’s probably the low-end couch there. You can imagine just the whole fine furniture industry concluding the average consumer, they don’t care about design. They just want cheap stuff. Then Ikea comes along and brings design to the masses. Do you think Ikea thinks, “Oh, consumers don’t appreciate good design.” No, they absolutely do appreciate good design. They’re not going to pay $20,000 for a well-designed couch. But they’ll pay $500 for a well-designed couch, maybe even $1,000.

Geoff Ralston [25:43] – It’s more of a balance for that sort of consumer, between cost consciousness and quality.

Keith Schacht [25:47] – I don’t think there’s something unique about the education space to conclude consumers don’t appreciate this. You create more value, you price it right, and you play the, you have to figure out the market dynamics to scale. Customers appreciate it. In our case, we started off focusing on teachers. The challenging thing in education, and this is also true of other industries, is how does the consumer know what’s better? How does the customer know what’s better? Teachers, our particular customers, they don’t have a lot of strong opinions about a better science resource would have the following attributes. So they’re not searching for better, necessarily, at least with a clear set of guidelines. But when they see it, they can recognize it. We hear all sorts of feedback from teachers, “Wow, I’ve never seen a lesson taught this way before. I don’t know what it is about it, there’s just something different.” You can see it in their testimonials that they’re sort of, they’re groping at their own identification of what it is. So they can tell it, even though they can’t put their finger on it. You can imagine a consumer, you make a healthier granola bar, and the average consumer doesn’t know, what is the right balance of nutrients to make something healthier? But they could notice how they feel in the long run by changing their diet.

Avichal Garg [27:04] – I think that’s actually a good example. Yeah, I mostly disagree with what you just said.

Keith Schacht [27:11] – Tell me more.

Avichal Garg [27:12] – I’ll give you examples, alright? Tesla started at the high end, right? Electric cars probably wouldn’t have happened had Tesla not priced it the way they did. There was a bunch of work that they had to do to come down. The iPhone kind of worked that way. A lot of technology cycles actually work with things that are more expensive, that are smaller in niche, and then they come down in price over time, because often there is a trade off between quality and cost. Now, in content creation, I think you could argue that the tooling has gotten so much better that the cost of production has come down dramatically. Therefore you don’t really have the same trade offs that you used to. With Mystery Science is a great example, right? Software tooling has gotten so good that you guys can have massive leverage and create really high-quality stuff and price it very competitively, right? It’s actually affordable for people, but it’s really, really high quality. But a lot of things don’t have, it is a trade off. To the extent that you have to make the trade off, that’s actually where things get hard, is the average person’s going to make, likely, a different trade off than the person building the product. That’s kind of where companies end up making the wrong trade off. The person building the product says, “Clearly if it cost 10% more but is 20% better or 30% better, that’s a worthwhile trade off.” That’s actually not the buying psychology of the consumer. On the margins,

Avichal Garg [28:24] – things get harder.

Keith Schacht [28:25] – But do you think that’s unique to education? Most consumer products-

Avichal Garg [28:28] – It’s unique to certain things, and I think the characteristic of education that makes it, there are two parts of education, which, it’s kind of like healthcare, that make it really hard. One is the long ROI cycle. It pays off in 20 years. Humans are just bad at that, right? That’s why healthcare has some of these challenges. The other is, your point, which is how do I know it’s better? What tells me that it’s better? If the payoff cycle was 20 years for granola, right, better food for you, nutrition’s a great example. People make all sorts of sub-optimal decisions in the short term when it’s clear what the right long-term decisions are. Most people just don’t make the right long-term trade off. It’s exactly this, is the payoff is really, really long, and evaluating quality in the short term is really tough.

Geoff Ralston [29:11] – You wouldn’t really argue that people don’t make quality trade offs, though, for food? They sure do. They might be wrong. They might be a little opaque. But they sure do make those trade offs all the time. That’s probably true in education, as well, and healthcare.

Avichal Garg [29:24] – How do you mean?

Geoff Ralston [29:25] – Well, people think certain granolas are better than other granolas and think certain berries are better than other berries.

Keith Schacht [29:30] – Totally.

Geoff Ralston [29:32] – And think-

Keith Schacht [29:33] – They’ll pay extra money for organic, for example, because they believe that-

Geoff Ralston [29:35] – Of course there’s, and part of that is that there are shorthand for quality, taste better.

Avichal Garg [29:43] – Of course.

Geoff Ralston [29:44] – A Mystery Science lesson tastes better than a dry textbook, without Doug actually doing it. So maybe there are ways for people to signal quality and to believe in quality.

Avichal Garg [29:54] – The way that we would evaluate quality here, making it fun and accessible, absolutely part of quality. But fundamentally, it’s about driving educational outcomes, right? If you didn’t learn the thing better, is it actually better quality? I’d argue if you didn’t actually learn the thing, it’s not better quality.

Geoff Ralston [30:13] – So we need shorthands, because the ultimate educational outcome, which means you’ll get a better job, it’s more satisfying, higher paying, etc. is too long a cycle. What schools use and parents by default then use is test results.

Avichal Garg [30:27] – That’s right.

Geoff Ralston [30:28] – I guess those are the outcomes that we have to use as a shorthand.

Keith Schacht [30:32] – I will argue that this is more of a symptom of a mis-measurement of better education. Since so much of a measure, the way that people evaluate whether a product is a better educational product is very often with some internal set of standards and heuristics that a district will use for purchasing, or it’s a standardized assessment that they’re going to measure at the end of fifth grade and eighth grade or whatnot. I personally question whether those are actually proxies for quality at all. Meanwhile, the thing that they ignore, which is the right proxy, is what’s the level of student engagement? Are student excited and eager to take the next step in that lesson or to understand this concept? That’s largely ignored by most people designing educational products, because they’re striving for some platonic ideal of raising a standardized assessment. They’re just missing it. When you understand that a child’s motivation is central to learning, one of our internal principles, you can’t teach a child something that they don’t care to understand. If they don’t want to understand it, no amount of trying to shove it down their throat will lead to any kind of learning outcome whatsoever. When you understand that, when you see motivation as central to learning, you get an immediate feedback loop. Students respond right away. You can tell whether they’re learning. Students want more of this. This idea that there’s a long cycle to measuring improvement is one of the sort of legacies

Keith Schacht [32:13] – of the educational system that we’re still working to shed, because we lack a real measure of learning outcomes.

Avichal Garg [32:20] – 100% agree with that. A huge part of it is a side effect of the way that we do the measurements and therefore the way that we measure quality, quote unquote, ends up skewing in a very particular way, which has all sorts of downstream consequences.

Keith Schacht [32:31] – The disconnect that people feel between, “Well, I made a product which is better according to this mistaken measure of quality, but gosh, it’s just not working in the market. I’m not getting the reception.” It’s not that the market doesn’t value quality. Question the assumption. It’s maybe what I thought was an indication of better quality is really not an indication of better quality.

Geoff Ralston [32:53] – Just to advance that conversation a little bit, something that I struggle with a lot is that the educational system that we’re stuck with today was building children, building outcomes for a world that doesn’t really exist anymore. The world has changed so much. One of the issues we have with measuring quality is, for what? There’s a lot of questions being posed as to what the value of an education actually is in actual longterm outcomes for children. How should we even think of that, and how would one advise an EdTech company who’s trying to build some platonic ideal in that world for what to do?

Avichal Garg [33:44] – It’s a great question. That is the core of the problem, is what are we actually building the educational system for, and towards what end? I don’t think there’s a great answer right now, honestly, and it’s as much a public policy question as it is a technology question. What are we actually trying to optimize for, and all that people are, for what it’s worth, what’s starting to happen is people are starting to at least realize that it’s broken. That whole generation of people that went through that system and ended up kind of misaligned for the skills that you actually need in the job market today are realizing that system is broken. But I don’t think there’s a, there’s not a cohesive plan that I’ve seen that says here’s how to actually fix it. I wish I knew the answer to that personally. But I don’t think there’s actually, I haven’t really seen a great solution to it, other than sort of very point-wise, “Oh, we should have more STEM education, we should get programming in earlier schools.” There’s sort of these point-wise solutions that are very tactical. But there isn’t a cohesive strategy that I’ve seen.

Geoff Ralston [34:50] – Programming’s an interesting one. Sorry, Keith, you wanted to say something. Go ahead.

Keith Schacht [34:55] – You absolutely have to have an opinion on that fundamental question. If you are building an education company, you have to have an opinion on what is the purpose of education? Because if it’s to create more of a certain kind of professional because of a career shortage, that will lead to a whole set of decisions and an assessment of progress. If it’s GDP growth or, there’s many I think mistaken views of what the purpose of education is. The fact that as a culture, we don’t have alignment on that, that speaks to the system that’s being pulled in so many different directions.

Avichal Garg [35:27] – I agree.

Keith Schacht [35:28] – We hold a very strong, unconventional view on it, and it’s that the ultimate purpose of education is that children grow up and live happy and productive lives. Even taking something like science, I think it is an absolute mistake to say that we should teach children more science because there’s a lot of STEM careers out there that need to be filled. Absolutely the mistaken approach. Many people hold that view.

Avichal Garg [35:53] – Why do you think it’s-

Keith Schacht [35:55] – It leads towards a bunch of wrong decisions.

Avichal Garg [35:56] – Why do you think it’s mistaken?

Keith Schacht [35:58] – It’s a great long conversation, when we’re trying to condense it. But I’ll give you a data point which will speak to it. If there was a young child who knew that he or she wanted to be a dancer, that’s the profession they knew they wanted to be in, there was no chance that they wanted to pursue a STEM career, that child, too, should learn science. It’s not that they should learn science because of a particular career outcome, because the implication of that is once I decide that’s not the career, than forget this! I don’t need this knowledge anymore. There is a fundamental baseline understanding of the world that you’re surrounded by, the way it works, and the habits of thinking that you form by figuring out how the world works. Science is this body of knowledge, this core knowledge about how the world works, and it’s a methodology in discovering these truths about the world. Every child should learn that, because it will enable them to be a happier and more successful dancer, just as it will a happier and more successful programmer.

Avichal Garg [36:58] – So where do you think the bounds of that kind of end? Is it computer science and programming? Is it math? Is it writing skills? Is it reading?

Keith Schacht [37:07] – That’s absolutely the right question. It is the harder question to answer, which is, so what does a core, fundamental, pedagogical, a pedagogy and curriculum consist of in order to maximize for as many children as possible, growing up and living happy and productive lives? There’s a lot of healthy debate that should be happening there. I wish that was the debate that was happening. I wish there was agreement on the outcome and the purpose of this all, because then we’d be having the right discussion, the industry would be having the right discussion.

Geoff Ralston [37:36] – I love that you make the point that people can live happier, more fulfilled lives by knowing science. I’m not sure that there’s a lot of evidence to back that up. But I would like to make a point and maybe start a discussion on the social case for certain kinds of education, because what certain types of education does bring to kids is critical thinking and the ability to ask hard questions and think skeptically about the world, which seems relevant today for all sort of reasons. Isn’t even a more powerful argument is that when we’re educating our kids to be productive citizens of the 21st century, and without an understanding of the scientific method and what science implies for everything, as an example, there’s more, right, how can you be that productive citizen? What do you guys think?

Avichal Garg [38:44] – That’s hard to argue with. Critical thinking skills, science, math, I would argue actually programming just as a framework to think logically and clearly about the world, writing. The ability to communicate effectively, these are all just foundational skills, regardless of what you end up doing. So no argument from me on that. The hard part is, can you be happy and productive if you don’t have a job? Just to bring it back to earth a little bit, is like, if I can’t feed my family, am I going to be happy and productive? That’s where the boundaries are, is like to what-

Geoff Ralston [39:17] – Wouldn’t you argue, as well, that if you think clearly and you have great communication skills, you’re better positioned to get a job in the world that’s being created all around us than otherwise?

Avichal Garg [39:32] – Maybe. I mean-

Keith Schacht [39:35] – I’d like to think so.

Avichal Garg [39:36] – I’d like to think so, as somebody who loves reading philosophy and science fiction and loves music and all these things, I would like to think so. But I think the edge cases are kind of the interesting ones. If you have an English degree and there’s a gross under-supply of data scientists, that feels like, could that person learn how to do Python and SQL and go get that job? Absolutely. They’re smart, they can certainly do that. But is that a missed opportunity? Is that person now in an awkward place, having taken $100,000 in debt when they probably should have just done, could have done something else, not probably, but may have? Those boundary conditions are the tough parts.

Keith Schacht [40:12] – You have to be really careful with the societal arguments. There was a subtle switch, where you’re saying but what about the societal? This is good for all of us, if people learn these skills. I agree with your restatement of it. It is, you’re right. It is good for every one of those individuals to learn those skills, because it will help them. But I don’t think we can justify this on the grounds of, but it’s good for society. I’ll give you an example. My wife was someone who was in AP math and science classes in high school. As a woman who was good at math, she got a lot of pressure to join the math team and be the female math representative for the school. It was really hard for her for a while, because her passion was history and the social sciences. But there was a pressure, because it would be good for society to have more female representation. She struggled with that for a little while. Ultimately, she pursued the things that made her more happy and fulfilled, not the thing that other people felt would be good for society because of some metrics that we need to hit. That as soon as you hold a societal standard, you very quickly start to make decisions which will be at odds with what any one individual would want for themselves, because it’s best for society.

Avichal Garg [41:31] – To probe in that a little bit more, the hardest trade offs are actually for the individual in those sort of edge case situations. It’s an interesting example. If your passion is in one direction and you could go really deep on that, but all the jobs are in the other direction, setting aside some baseline level of scientific method and math and you’re going to a good college, that’s kind of an interesting and hard individual trade off. That starts to get at actually what is the purpose and what are you optimizing for, as a system? Because in a lot of those cases, there are, you can look at, let’s say unemployment rates for people who have a college degree, and it’s actually relatively low, right? It’s below five percent. It’s really whether or not you have that college degree that changes whether or not you’re employed. But there are a lot of nuances to that. There’s underemployment, there’s what kind of salary growth potential do you have, and I think I take a little bit more of a pragmatic approach of like, probably more people should be getting computer science degrees and learning data science. That’s better-

Keith Schacht [42:34] – If more people want to be given computer science degrees.

Avichal Garg [42:36] – But it’s a critical and important part of being happy and fulfilled that you have a job that pays you well, and you can feed your family, and you feel comfortable. That’s an important part, above and beyond your talent choice.

Geoff Ralston [42:48] – It’s a whole separate conversation, whether you believe that having a job is a critical part of happiness. Because, and it’s easy for those of us who have the sorts of jobs that, frankly, the people in this room have, where we’re the opposite of wage slaves, which is what most people are. That’s a different conversation. But it is interesting to talk about the societal goals of education. Public education in the United States was for the benefit of society, as opposed to the individuals. I don’t think people necessarily knew that getting people ready to be good factory workers was good for them. It was more about, we need factory workers, so we better educate them sufficiently so they can be good. Let’s bring the conversation back to education technology. My hope is that educational technology helps both individuals and society achieve outcomes that we, in this room and in the country, will be happy with. What do you guys, today, recommend when you meet an EdTech founder? I suspect you probably say different things. Maybe you can each in turn say, if someone told you about their EdTech startup, it’s a hard general question to answer, but where do you guys sort of go with your advice? Do you, I’ll start with you, Avichal, do you sort of channel your 2011 self, or do you talk about sort of how the world’s changing and the opportunities that are thus created?

Avichal Garg [44:27] – I try to channel Keith a little bit. I ask them what their motivations are and try to do what’s good for the individual, cause I don’t think there’s a right answer. There are a lot of reasons that people might want to start an education company. The foundational thing is really what is your motivation, and why are you doing it? Especially in Silicon Valley, there’s too much of a lens of, if it’s not a billion dollar outcome or if it’s not a 10 billion dollar exit, it doesn’t matter. There are lots of motivations that people may have. The first step is what’s your motivation? There are lots of great businesses that people should start and could start that would have very, very meaningful impact on people’s lives and their communities. They’ll be 50 million dollar businesses or hundred million dollar businesses. That’s fantastic. People should go do those things. But if that’s kind of the zone that you’re in and that’s your motivation and that’s what you want to do, then it comes down to, what are the right tactics to executing on that, and how do you stay really smart about it? Because the worst possible scenario would be, you had this great thing, it was actually working, but for example, tactically, you just ended up raising money in the wrong way. Now you’ve kind of gutted your company because you don’t actually control it, you had these options or mis-priced, nobody can actually make money by being at your company, so all your best people leave. You enter this bad death spiral

Avichal Garg [45:35] – if you grew your company in the wrong way. The fundamental thing is what is your motivation, why do you want to do it? Okay, let’s figure out what the best way to make that happen is. There are tons and tons of great education and EdTech businesses that are all on that spectrum.

Geoff Ralston [45:49] – Sounds not too different than any other startup.

Avichal Garg [45:51] – Any other startup, in that way. Tactically, the specific thing that I say about education is, I try to get people to really double down in their understanding of what is the actual thing you want to accomplish. You have your motivations, and if you’re successful, a certain thing will happen in the world. Is the path that you’re on the best way to necessarily make that thing happen in the world? Are you better off, for example, if you’re really into computer science education, there are lots of ways to make that happen that are not necessarily an education product, right? Are you actually better off trying to do job retraining? Are you actually better off doing it outside of school? Are you better off doing it inside of school? Is it actually at the high school level or college level? Are you going to target sub-populations? There’s so many ways to actually go about making that happen that are sort of adjacent to EdTech. This kind of gets at is like what is EdTech exactly is an open question. They’re kind of like EdTech heavy and EdTech light and EdTech adjacent.

Geoff Ralston [46:43] – We’ve really sort of skirted over the fact that, actually Keith mentioned in the beginning, there’s so many aspects to EdTech. You can slice and dice in many ways, from the homeschool market to the elementary school market to the high school and post-secondary market to training sorts of markets. At YC, we funded lots of schools that train folks for-

Avichal Garg [47:07] – Insight Data Fellows is a great example.

Geoff Ralston [47:08] – Insight Data, Lambda School is training people to be programmers. It’s actually a pretty broad field. It seems like the opportunity has only grown in what we’re broadly calling EdTech, even in the United States, Asia completely aside, right?

Avichal Garg [47:26] – Totally. Actually those things that are a little bit more on the fringes, to me are some of the most interesting, because they start to sort of redefine what it means to get an education, or they redefine outside the bounds of the sort of public education, real estate funded, tax funded sort of model and start to explore what it means to actually get an education.

Geoff Ralston [47:47] – And stay educated.

Avichal Garg [47:48] – And stay educated and stay relevant, exactly. Those kinds of models are actually where there’s a lot of potential innovation, cause they’re free from a lot of the constraints that exist in these models that have existed for 50 or 100 years. That freedom breeds a lot of innovation.

Keith Schacht [48:05] – I agree with most of what Avichal said, particularly with regards to giving someone advice on picking a business they’re going to start in. Most of my focus, as well, is on what are you going to be so excited to work on that you’re not going to give up after two, three, four years? Cause it’s probably going to take you 10 years. That kind of heuristic, when you’re picking a business, is more important than almost anything else. But specific to the education space, there’s one, the most contrarian thing that I would share with someone is that the education space, I like to say that we’re a learning company, not an education company. I’m very interested in learning and how to improve learning. When you think of yourself as a learning company, learning is a field that’s in its infancy right now. It’s deceptive, because it’s an old industry, so it feels like we’ve figured all this out. It’s just like some efficiency gains that are left, and things like that. Learning is much more like genetic engineering or in figuring out how to arbitrarily sequence DNA and inject it into an organism. The level of understanding we have about that, we’re still poking this black box of genetic engineering and trying to make stuff happen for the first time. That’s the state of learning. Fundamentally, even though science has progressed for so long, the science of the mind is still at such a beginning. You take a field like physics, and we know so much about the furthest corners

Keith Schacht [49:31] – of the universe and the subatomic level, whereas our level of understanding of psychology is at the level of physics when it was, maybe the gods are angry and they’re shaking the ground, and that’s why there’s an earthquake. We’re still not that far beyond Freud, where it’s like maybe you went into this profession because your parents had a particular relationship with you, and they steered you in that direction. It’s very much like opening a computer and noticing that when you press this sequence of buttons, this part heats up. That’s our level of understanding of the brain, is like we image this thing, and we notice where the blood flows and try and draw these conclusions about how it’s working. Anyone who wants to start a company in the learning space, to the extent that you really think, focus on understanding the science in learning, it’s a wide-open territory. We are actually still discovering profound truths about even something as simple as space repetition and the power of that, or the role of concept formation and what’s the difference between being able to recite a fact and actually holding that fact as knowledge? We’re at the beginning of a science. To the extent you figure out anything new, there’s huge business implications of that.

Geoff Ralston [50:44] – Are you optimistic, Keith, that we’ll figure it out and that we’ll see a transformation in American education over the next five, 10, 15, 20 years, that will actually make for better outcomes for individuals and society? Then maybe, Avichal, if you could answer the same question afterwards, we could sort of close on that note.

Keith Schacht [51:10] – Absolutely. I mean, I will say I’m long on the human race, right? We are very good at figuring things out and mastering sciences and plumbing them to new depths. We will do that for learning and for pedagogy, as we have for all of the other sciences. There is this industry and this infrastructure, education infrastructure, that will crumble and go away and be replaced by another version and another version. What’s so exciting about the internet, in the internet generally and mobile in particular, is these disruptive forces give you these opportunities to rebuild the whole stack from the ground up. Most likely the business opportunities in the learning space will be outside of the existing stack. The fact that you can go straight to teachers and not have to go through an approval process is dramatically changing things. The fact that there’s an internet and the cost structure fundamentally changes.

Geoff Ralston [52:06] – To parents, too.

Keith Schacht [52:07] – So the science is wide open, the industry is going to be transformed multiple times in the next decade, and we will take learning to the next level.

Avichal Garg [52:15] – Hard to argue with not being optimistic about the human race. Yeah, very optimistic about-

Geoff Ralston [52:21] – Not that hard, actually.

Avichal Garg [52:23] – No, I’m an optimist.

Keith Schacht [52:25] – Many people wouldn’t.

Avichal Garg [52:26] – As entrepreneurs, you have to be an optimist, right? You can’t imagine you’re going to change the world for the better if you aren’t-

Geoff Ralston [52:34] – It’s almost the definition of an entrepreneur is an optimist. You know it’s impossible and you’re still going to go-

Avichal Garg [52:39] – How do you decouple that, right? So as an entrepreneur, I think it’s hard to separate that. All my beliefs are tinged by that personality trait. But what I would say is I am very optimistic about the experiments that are happening by reaching teachers in new ways, reaching parents in new ways, platforms like the iPhone or Android or iPad, Chromebook, that sort of allow people to interact with information in novel and new and interesting ways. Things that are happening on the fringes, things that are education but don’t look like education. Insight is a good example. People sort of exploring the internet on their own and kind of going down a rabbit hole of open source or YouTube as a platform, with Khan Academy. There are all these really interesting things happening on the fringes, where some of them look like education, some of them look more like learning, and some of them look like job training. So they all kind of have different flavors. I’m less optimistic about the the likelihood that we dramatically change the state of public education in the United States anytime soon, because there’re just so many entrenched components to that, from funding to the psychology that Keith was talking about, where for the first couple years, it’s really on the parent, and then there’s a handoff. There’s so many cultural and governmental and public policy kind of things kind of rolled into that. Unbundling that is going to take a long time. There are a lot of entrepreneurs

Avichal Garg [54:04] – kind of on the fringes that are doing really interesting things. That’s often how disruption happens, right? It kind of comes from the side a little bit, in very unexpected ways.

Geoff Ralston [54:11] – The timing can surprise you. Maybe we can leave on a slightly more optimistic note, which is that I think even in public education, we know a lot of what the solutions ought to be. We’re kind of in a unique position, where there’s a lot of structural reasons it’s hard to close the achievement gap and to overall improve outcomes, but we do know a lot of the outcomes. Since the stakes couldn’t be higher, I think that we have good reason to believe that that knowledge will eventually matter and that things will get better. I want to thank you both for podcasting with me this afternoon. It’s been a great conversation. I’ve enjoyed it a lot. Thanks.

Avichal Garg [54:56] – Likewise.

Keith Schacht [54:57] – Thanks.

Craig Cannon [54:57] – All right, thanks for listening. So as always, you can find the transcript and video at And if you have a second, it would be awesome to give us a rating and review wherever you find your podcast. See you next time.


  • Y Combinator

    Y Combinator created a new model for funding early stage startups. Twice a year we invest a small amount of money ($150k) in a large number of startups (recently 200). The startups move to Silicon