A Gradual, Evolutionary Approach to Disrupting Education

Interview with 17zuoye Cofounder Dun Xiao

Dun Xiao, 17zuoye Cofounder

Dun Xiao’s pursuit of a quality education has taken him around the world, from Cambridge University to MIT and Harvard, where he studied electrical and information engineering. Armed with a lifelong passion for education and his technical background, Xiao Dun was inspired to found 17zuoye.com, (a.k.a. Homework Together), the largest online homework platform for K12 students in China.

And so today, we explore Chinese Edtech — a market of particular significance in China given its traditional reverence for learning and the lucrative commercial opportunity this yields. With Dun, we explore key topics and trends, including: how do you segment the Edtech market in China (e.g. by learning groups, product and delivery type?) How do you look at future of education… does this involve building disruptive products that can be applied immediately (e.g. Alt School or Minerva in the US), or should it be more of an evolutionary process where technology is used to complement existing models? What drives key stakeholders in education ecosystem, and why is it relatively more difficult to scale an education business that involves working with teachers, students, parents, school systems, and the government? What is the role of data in education? How do you apply customized learning solutions onto the existing education system? And lastly, how do you get teachers to buy in on using new teaching solution and onboard them efficiently.

Listen to the podcast on SoundCloud (here)

Transcribed by Shaolong Lin and Tyler Xie, edited by Yvonne Yan


Meaning of Education

Adam: Welcome Dun. Could you start by telling us a bit more about your story, and how you ended up founding 17zuoye?

Dun Xiao: Hi Adam, thank you it’s an honor to be here. I’m Dun, co-founder of 17zuoye, also known as “homework together”. Our company has had over 5 years of history and are pretty much still a start-up. I was born in Beijing, and I was in Beijing for 16 years. Then I went abroad, to the U.K. and the U.S. to study electrical and information engineering. After graduation, my first job was with UBS in London, doing structured products trading, devising trading strategies for them based on data analytics. Then I was involved briefly with the organization of Beijing Olympic games. That had a big impact on my view of the world in achieving and making something that is meaningful.

When I thought back to my life about things that had an impact on me, I came up with this word, education, which I thought had the biggest impact on me. Probably it will still have the biggest impact on not just me, but also people, and the future of mankind. I came from a technology background and I think technology is going to be one of the drivers for the growth of our society. It can be applied to pretty much any industry, but applying it to education industry would be meaningful. That’s why I started my first start-up in London in 2008. It was a very simple idea, combining contents with technology, so we were among the first developers to doing educational apps. Then in 2011, I came back to Beijing, and I focused on the Chinese market, and co-founded 17zuoye. That’s basically how we came to existence, and we are still here, after five and a half years. We are very happy about that.


How Technology Transforms Education

Adam: You mention that education is very important and valuable, and I completely agree. That’s part of the reason why we are running this podcast, sharing Chinese innovation and technology trends with a western audience. But help us understand the ed-tech market. It’s a very vast sector, and there’s a so called Massive Open Online Courses (MOOCs) like Coursera, there’s a skill-based course work like Udemy, there are AR/VR applications. They even have a very different type of school experience called Alt school. So looking at the Chinese market, can you tell us a little bit more about education technology and how it’s broken out.

Dun Xiao: Education is a big word. It’s also a big industry. There are many ways to break it out. First maybe by age. There’s a pre-school education from age 0 to 6, there’s a K12 education from 6 to 18, and then university and graduate school education as well, from 18 to 22 or 25. There’s professional adult development education, from 25 upwards. Perhaps there’s a 60 and above education for people who are retired. That’s one way to divide it.

The other way you’ve mentioned is through the delivery mechanism, and I think in this sense there are so many models in China. New models are coming out almost every year, and using new technologies. When we started, we were purely PC based. Now we are 80 percent plus mobile. We are also looking into things like the VR and AR, and we are already doing a lot of life broadcasting. There’s also obviously MOOC. My worldview in this sense is that the different models in ed-tech are actually kind of like BAT. They’re all different worldviews. Some use very narrow settings to acquire as many users as possible, like Baidu. Some just go through the transactions, and acquire users through GMV like Alibaba. Some try to build a community of users that are very sticky, like Tencent. Also there’s a Xiaomi or Apple worldview, but it’s not so prevalent at the moment. I think the hardware devices like mobile and PC are not perfect for learning. I think there’s opportunity here to make the next generation learning device. Also you’ve mentioned that there are some innovative schools. That’s happening in China as well, combining technology and education at the micro school level.

Obviously there are other ways to break it down as well. Our company focuses on K-12. The K-12 market is already a big and somewhat complicated market itself, there are lots of different people doing different things. Some are purely B2B (Business to Business) that work with government and schools and get their revenues that way. Our company is primarily B2C (Business to Customer) but also working with schools as well. That makes us quite unique.

There are many models, and I think together we are trying to solve this problem of how to educate the next generation of human beings. Because of the advancement of technology, I believe technology is going to have a bigger impact on our lives in the next decade or two decades. A lot of jobs are going to be eliminated and new jobs being created. We need people who have these skills to do these jobs, and also people need skills to be happy in this next generation. We need to work on the education industry now to make sure we grow this kind of person.


What’s Unique About 17zuoye?

Adam: Thanks for that. Clearly you’ve been very thoughtful about the entire universe of all the different institutions and also different delivery mechanisms that you mentioned. Now can you tell us more about how 17zuoye fits in. What is the product, who is involved, and how you deliver that service?

Dun Xiao: Maybe I would just come back to the last question a little bit. There’s perhaps another way to divide the education services or companies. One way is basically to create a product that is ideal, and then you apply it to the industry right now. There are some of these very notable products, like some schools for example, Alt school, Minerva. Another way is to think about the future and to work on the present, and then make the present more and more like the future, but make it more of an evolution rather than a revolution. For the former path, there’s a big trend in China as well, like the introduction of STEM education, introduction of science, technology, art and mathematics, which is a kind of cross subject, trying to train students creativity, leadership and problem solving skills. For the later model, you need to actually work with the system. For K-12, we are more like the conservative view, the evolution view. We think the education system needs to be changed, the way we learn needs to be upgraded. But it can only happen from the present. So we have decided to take the hard way and work with the current system.

In K-12, the stakeholders are students, teachers, parents and perhaps the government as well. We are now working with all these people and I think we are the first company in the world to have come up with a model to work with these people and produce in-school services for free, and in return we get users and data. Through this, we can offer premium services to the end users, the students and parents. Our business model is charging the students and their parents on premium services, including contents and courses. I think so far it’s been proven to be a viable business model, and it is a very heavy approach.

Take Uber for example, you are dealing with two people, the passenger and the driver. For Amazon, you are dealing with the seller and the buyer. But that’s it. It’s very rare that you are dealing with more than two business groups. But our situation is more complicated. Without teachers there’s no retention, and without students there’s no learning data, and without parents it’s hard to monetize. We also want the support of schools and government to make this transformational model stable and scalable. That’s kind of what we are dealing with here.

Most specifically, we found this so-called entry point, “homework,” which is already something that automatically connects the core stakeholders of this market, teachers, students and parents. The teachers need to initiate the homework, students do the homework, and teachers and parents check the homework. The school also cares about what students do in the homework.

We started with English listening and speaking homework back in 2011 because there was no good way to train student’s speaking and listening. I went to the U.K. when I was 16. I remember very vividly, that the first morning when I got up, a British student asked me, “how are you?”. I just said, “pardon?”. He said, “How are you”. And I said, “sorry?”. And he said, “How are you”. Finally I understood he was asking me “how are you”, and I said “Fine thank you and you?” I’d been actually studying English for 10 years before that, and I was once considered one of the better students in my English class. But I couldn’t communicate well via spoken language. There are more than 100 million primary school student in China who either learn English since the age of 6 or 8. In theory, they should be very good at English. But their English teachers actually sometimes cannot speak English!

Also there’s a very low teacher-to-student ratio, on average 100:1. It’s very hard for the teachers to help the students with their listening, speaking and pronunciation. There’s really no way for students to practice speaking English with pieces of paper for listening and spoken English. That’s how we started, using technology to give a feedback instantaneously on students’ pronunciation. We can then have the students put in an input, like “welcome to The Harbinger podcast”. The students can repeat, and we can than score the pronunciation, then give feedbacks to students so that they can try again. The students are happy with this process, although it’s not perfect, and the teachers are obviously happy as it’s better than nothing. It makes the students engaged and interested as well.

From there we expanded to other types of the homework, including mathematics, Chinese literature, and now we are going to junior and senior high schools, with subjects like physics, chemistry, biology, geography, politics. Apart from homework, we offer other products as well. For example, teachers can exchange materials, prepare classes, have discussions with each other. Or the students can do some self-learning or additional learning after homework, while the parents can find resources and communicate with teachers. At the very bottom layer we have a homework platform, that’s kind of our main service up to today. We also have other contents or education services for these three types of users on top of homework. At the top level, our worldview is very much like Tencent: essentially a community of teachers, students and parents who are organized by classes. It’s the community between teachers and teachers, students and students, parents and parents, but also cross these users. Teachers can communicate with parents, parents can communicate with students and with each other.


Freemium Business Model: Individualized Coaching

Adam: Sounds very valuable, especially in China. In terms of your business model, you mention that there’s a kind of freemium piece to it. The homework application is free, but beyond that, some of the premium services are where you actually monetize. Can you talk us through this a little bit?

Dun Xiao: Homework is a very frequent behavior in China. There are 100 million primary school students, another 100 million high school students; pretty much they do homework everyday. That’s about half an hour of homework in primary school on average and about 2 hours of homework on average for high school. Through this we get lots of data. Before, there were lots of data on pieces of paper and they are not available there for analysis. We actually think these data very valuable, so in 2016 Q1 we had over 100 million questions solved on our platform. In Q1 this year we have over 12 billion. It was almost 100 times growth in one year. As the students solve these questions, I think they grow as well. From our perspective, we can now analyze those data and learn more about students’ learning and teaching objectives.

The reason this strategy is valuable is that we have gained a much better understanding of the content, pedagogy and teaching, so that we can recommend to students and parents lots of things. For example, you need to learn certain things in this semester, and teachers assign some homework. But we’ve found out that there are certain things that teachers miss. Then we can simply recommend to students and parents these additional things and help them to fill in the gaps.

Also we can find that for example, this student is very struggling. He is doing homework very slowly, sleeping very late while not mastering the materials at all. We must help this student by giving less challenging materials and guiding him through this course. Or we might find the opposite. Some students might totally find it easy, in this case we have to give them more challenges. One thing that we did in the old education system was to use industrial revolution production model to treat humans like standardized goods for them to work in standardized factory positions. But in fact, everyone is different. This idea coaching or teaching according to student’s’ innate skills or characters has never been done properly. Technology and data are very good driver for this. To answer your question, our premium services are individualized learning solutions recommended to these users.

Adam: I see the value of that, and it seems to me like customized learning. Customized learning in a vacuum could make sense, but if you are applying it to existing schools, existing structures, teachers and typical curriculum, how do you reconcile customized learning with existing standard curriculum that they have to go through?

Dun Xiao: That’s a great question. That very much ties in with our idea of whether doing the future now or applying to the present. We started this product basically in sync with the teacher’s curriculum. We look at the textbooks used by teachers in schools in China, and we produce contents based on this textbook. At the bottom of content production, there’s this so called internally recorded goal, system, which is a vast universe of learning objectives including knowledge, skills, aptitudes, processes, emotions, values and so called identity. These are things that we think need to be learnt.

For knowledge, they are interconnected. A huge map of interconnected learning objectives. Some knowledge points in mathematics, for example, could be applicable to physics, chemistry or even Chinese as well. Then whatever textbook the teacher is using, we first push to the teacher knowledge and skills based on the textbook.

Take English as an example, perhaps this teacher is teaching this topic: my family. The vocabulary knowledge in this unit would be daughter, father, mother; and we’ll have lots of contents based on this. Maybe there are some questions, for example we have a picture of a lady, and there are some choices: brother, father or mother. Maybe there is a pronunciation of the word father and a picture of a man. Maybe there are some other types of open-ended questions like interview your family and produce an essay, or do a PowerPoint to introduce your neighbor’s family. Or have a discussion of whether the pet is a family member. All these contents that have this tag (my family). Eventually for the homework, everybody has the same homework at the beginning; as the students are learning, we have a better understanding of how well they’re mastering the content.

Now take vocabulary as an example. There’s listening, comprehension, spelling, writing, etc. As the student is learning and generating data for us, we can pinpoint each student’s problem. We can also pinpoint the class problem. Then we can make recommendations to the teachers. Through their usage of our application, the system learns again to determine whether our recommendation is useful for the teacher. Then this evolution happens. As teachers and students use our system more and more, our system gets smarter and smarter. It’s not just for this class. There are actually a lot of similarities between how people learn and how teachers instruct. A lot of the experiences are then consolidated into our algorithms and can be applied into many situations. That’s basically how we evolve. And we will get more and more data/materials, and we will cover more and more settings in learning to make more accurate assessment and learning recommendations.


Making a Teacher’s Job Easier

Adam: Right. I’m just thinking about the education ecosystem you mentioned earlier. The service is straightforward for students. It’s a more interactive and playful approach to learning. For parents, it is definitely something new. And for the teachers, clearly, it’s valuable. Most of the effort is going to help them use the product because a lot of control points lie with them and the value creation here is for them to be a better teacher.

A clever friend of mine (Jim Liu) once told me that for a lot of products in the education space, the value there is to help raise the baseline ability of teachers. Because nothing’s going to replace a great teacher. But in situations where you want to improve the baseline, that’s when you bring those products and services.

So let’s focus on those teachers. Tell us a bit more how you onboard those teachers, how you make sure to maximize the value of products with these teachers.

Dun: We think that teachers’ need from those educational products is to make their job easier because they are already very busy. There are so many things we can improve on teaching through combining with technologies, but teachers have not been using a lot of those technologies. If you go to Chinese K12 schools, you will be shocked by the degree of using technologies. That’s not to say that they don’t have the money or resources to bring in those technologies. The primary reason is that there is an added burden from learning and getting used to those technologies. They are very busy already and don’t have the time for learning and training. So, one thing we thought about really hard is how we can help teachers reduce or lift the burden.

Luckily for us, there are six processes in homework: homework preparing and assigning, then student doing and submitting the homework, lastly grading and providing feedback by teachers. Teachers are involved in five out of these six processes. So, we are trying to make those processes easier for teachers, especially for marking homework. That’s something teachers spend 10 hours a week on average. And that’s usually a repetitive and manual work. For example, there are multiple choice questions, which are easy to mark but very manual and repetitive when you have 100 students. And we see that this can be done by technologies rather than by humans.

And after grading, there weren’t easy ways of analyzing the results before. Now we can do the grading and analysis of the results with a click of a button. Even more, we now can grade listening and speaking in English, essays and some math questions. And now we’re working on those unstructured questions, for example, make a PowerPoint about your family, which is very difficult for our system to grade. In that case, we will facilitate student discussions with teachers for suggestions. And we’re trying to make the computer to do more grading. This is the core value of our product.

Also in other processes like preparing the content of homework, our product helps teachers prepare the content in line with the curriculum by facilitating UGC, user generated content. So teachers are motivated to use the content provided and create new content. And assigning and submitting homework are automated. We also enable teachers to provide feedback to parents individually as well, facilitating teacher-parent communication.

But homework is only one part of education. Our product is already very good. Actually, it is one of the best products in homework in this niche market, but it is still very hard to incentivize teachers to use it. And it is not the teacher’s’ job to look for and try products and technologies in the market. Even if it just takes five minutes to understand our product and probably 10 minutes to experience it, some motivated teachers may do it, but most of the teachers won’t as this is not their job.

So we have a team of offline operation, or teacher trainers basically, to educate teachers to use our products. They are a team of 500 locally based trainers across around 100 cities in the country. They go to schools to explain the product to teachers, a bit like sales, except they don’t sell anything as using the product is free of charge. Sometimes they meet teachers after school and ask them what are the problems they are facing in marking homework, how we can help them to lift the burden and improve the quality of teaching? How can we make students more engaged, make learning more efficient and individualized? And things like that.

And this is an ongoing process. When we have major updates, we will explain them again. For example, now we have a teacher’s forum, so we go talk to teachers again to encourage them to talk to each other and share ideas in the forum. Also now we have a survey function, which is helpful in the teacher-parent communication. For example, teachers can ask parents for suggestions on what books they would like the school library to have for students to read. Teachers can do things like that through surveys. Then the teachers are constantly in touch with us. Essentially, I think education is always about service. It’s not just a product or content. So using the survey can really help teachers improve the quality of their teaching, which also is one of the reasons they are using the product.


Disrupting the Existing Schools System?

Adam: Got it. You mentioned a lot of points here. You have created a great experience for parents and students, and we can see from the reviews that it has been very positive. The hardest part is getting the teachers onboard. And you have been doing great via those approaches you mentioned. You also seem to have a good government support; maybe we can look into that a little bit separately.

So my question here is that you applied an innovative business model in this existing traditional system, do you think you can eventually separate it out and create a more disruptive business model on its own. People are using this as a solution for better teaching. Can you just create a separate curriculum or even a separate school altogether?

Dun: Well, I think there is a possibility for doing that for after school, where students are kind of free in terms of time. Most of the K-12 students in China spend a lot of time in school. They go to school at 8 am, finish at 3 pm for primary school students, 5 pm in lower secondary school students. In upper secondary school, year 10 or 11, they may finish at 7 pm. In year 12, which is the year of preparing for the university entry exam, students may finish school at midnight. And they also spend a lot of time in school even after school time based on teachers’ instruction.

So to develop curriculum, we still need to work with teachers and schools. If we were to create something entirely outside of the system, the content could be better than the one in the system, from the learning perspective. But it would be very difficult for teachers to adopt. Because it would be different from the textbooks required by the governing body of education. Of course, we can try to link the content created to those textbooks, but still, it would be difficult for schools and teachers adopt it straight away. But I think we’re working on this direction through an evolutionary approach.

Adam: So maybe it will happen in the future. But right now, from what I’m hearing here, it seems to be more practical to work with the existing system.

Dun: Yea, that’s my understanding of doing the educational business. You know, in China, the first educational entrepreneur is Confucius, over 2,000 years ago, who had 3,000 students. His ideas are still impactful in China, and in the U.S. as well. There’s that kind of an ideal education, which is passing on the wisdom through generations. Some of the wisdoms may be outdated or even proved to be wrong, but the right ones should be passed on. I think the schools and teachers are not just a teaching agent. School is more like a reflection of the bigger society. Those values, things we work for and pretty much everything are happening in society also happen in schools. So schools are the environment for students to learn about society, especially in China, with those authorities involved. And I think this is crucial, is not just about knowledge acquisition. And that’s kind of the rationale behind our revolutionary approach.

We will stick with that until there is a possibility to replace those teachers and schools, which I don’t think will happen in the next 20 years. Even with the advancement of artificial intelligence, I think teacher will be one of those last jobs to be replaced. Once again, it’s a service, but it needs a lot of human touches. We consider our role at the moment, is more like an assistant coach. Let’s say tennis. There is a coach who teaches you how to serve, how to hit the back end. There is also an assistant coach who practices with you, trains you when you’re not doing so well, tries to encourage you when you’re down and manages your expectation when you’re too cocky. So that’s the kind of role of an assistant coach that spends more time with you than the coach. But the coach is still essential.


Chinese Market vs. Overseas

Adam: So one more ask of wisdom and coaching from you, because this podcast is about bringing Chinese innovation to the west and the rest of the world. Can you tell us a bit more about how you see the Chinese market versus others, and the variances when it comes to business models or technology developments?

Dun: I think we can learn from each other a lot. In terms of learning, we have a slightly different approach. In China, because we have this university entry exam, which can change the life of a student and the family, students take learning in K-12 very seriously. We place more emphasis on how the students master knowledge and concepts, which needs to be changed to some extent. But this also lets students do quite well in things like mathematics. Chinese students are quite fluent in terms of playing with mathematic concepts. There are also things the US students are doing very well. For example, communications skills, leadership skills and critical thinking, are traits essential for the next generations.

In the education industry, I believe that content technologies are at the core. The U.S. companies are more advanced. They spend more time and efforts on creating content with a very good quality. You know, they’re getting experts from all kinds of fields to produce one piece of content may only be used for 10 minutes. But that 10 minutes may influence many students’ lives and future, so they take it very seriously. And the intellectual property of content is better protected in the U.S. as well. These are things we’re really excited about. Because before, the education you get is related to when and where you’re born. Now with the technologies, you can travel through space and time to get an education anywhere anytime for anything. You can basically get the best contents from the world wherever you are whenever you want. And this is something we can definitely learn from the U.S., to produce some great content and put them on our platform.

Another thing is technology. We are working on a range of technologies to recognize handwriting and automatically translate to a language that computers can understand, and to mark and give feedback. And we’re developing chatbots to have conversations with students, rather than just to provide pronunciation in language teaching. We’re working on something called TTS, which is text-to-speech, generating speech from text, so we don’t need to have humans to record the things we say. We’re also developing robots to help translate, help students to understand, also generate questions for us. One of our robots took part the university entry exam of mathematics this year and got 105 out of 150. And actually those questions were too difficult linguistically for computers to understand, so our robot failed in Chinese but did quite well in the math part. So there are many things we’re working on. In this sense, we can learn from the U.S. in technologies like big data, artificial intelligence and deep learning. And there is a very bright future for the application of these technologies in China.

And the U.S, companies tend to be more focused on content providing. We’re working with a content provider specializing in the picture books for 20 years in the US. Their graphic contents are really good. Some companies are specialized in some particular content technologies, and they can build a great business by just selling their technologies to other businesses. While in China, companies are less specialized and doing more diverse things. Because it’s a big market. And we want to do a lot of things and reach a lot of users, and may not so specialized in particular areas. I think this part we can work with each other to get better results.

Adam: Thanks for that, Dun. Clearly, there is a lot potential for knowledge sharing between East and West. And, again, congratulations on the very successful company you built. It’s a very innovative business model. Also, it’s very honorable that you’re applying your efforts into building products valuable for other people, based on your personal belief about education and its value.