"I never expected that the small idea I came up with in 1998 would result in such a big business," said Yang.

A veteran of the education industry, Yang has experienced the industry's transformation from offline to online. "It's just like when I used a Nokia back in the days before the iPhone. I couldn't imagine how to type on the phone without use of a physical keypad. But since the iPhone, I no longer know how to use a Nokia. This is exactly how I feel now," Yang said.

In 2015, iTutorGroup completed its USD 200 million Series C round. Its investors included Government of Singapore Investment Corp., China-Russia Development Fund, Goldman Sachs and Silverlink Capital.

This record-breaking financing occurred on the eve of China's education market reform in 2018, now recognized as the year in which Chinese education enterprises made intensive entries into the capital market, launching IPOs and listing on stock exchanges in the US.

Focusing on the adult online English teaching market, iTutorGroup is now carrying out its last round of financing in readiness of next year’s IPO. Eric Yang says iTutorGroup will use the money to invest in AI, speed up development of its learning platform and intensify investment in youth-related projects.

Is a traffic mindset really important?

The number of China-based education companies listed on US stock exchanges took a dramatic leap in the first half of this year with IPO launches of Sunlands Online Education Group (NYSE:STG), OneSmart International Education Group Ltd. (NYSE:ONE), Ambow Education Holding Ltd (NYSE:AMBO) and Puxin Ltd (NEW).

Online education is a good fit for China’s first-tier market due to its mature business model and predictable market share.

In January, DaDa finalized its USD 100 million Series C round co-led by TAL Education and Tiger Fund. At the end of June, VIPKID completed a USD 500 million strategic financing co-led by Coatue and Tencent.

iTutorGroup has initiated a new round of Pre-IPO financing ahead of its public listing next year in the U.S. or Hong Kong.

Due to the massive capital flows into the education market, the industry as a whole is suffering losses. Public data shows that since 51Talk unveiled its financial report after its 2016 IPO, the company has operated at a loss, while Bloomberg has revealed VIPKID’s losses amounted to RMB 1.162 billion in 2017 and will reach RMB 1.8 billion this year.

iTutorGroup is yet to unveil its profit status. But according to Eric Yang, the company realized operating revenue of USD 350 million in 2017, and is seeing a positive cash flow.

"Some companies constantly raise funds then use the money to compete for teachers, students and market share,” Yang said. “In my opinion, this is not the correct guiding principle for the industry.”

From his point of view, acquiring students through massive spending alone is inappropriate due to the limited online effects of the internet economy in the education industry.

One more cruel truth is that it costs near to nothing for online education users to transfer from one platform to another. This is similar to the medical industry, in which if a patient fails to see expected results with one hospital, he/she would transfer to another. Similarly, if a student fails to see positive learning results on one platform, he/she would turn to another or demand a refund.

"From an Internet perspective, acquiring large numbers of users can form a barrier to new entrants. For example, WeChat fended off MiChat through mass user acquisition; the more friends you have on Facebook, the more unlikely you will turn to other social media. However, education is not a winner-takes-all industry; in the education world, no specific company can forever rule the pack.

Previous business models such as traffic generation from free content or free tools, such as test pools, have proved failures. Business models based on free traffic are unable to form a complete closed-loop business once payment is involved.

If customers are used to getting free products on a platform, the willingness to pay is low. Products that are largely and quickly duplicated are deemed as free in the Internet era, and with near zero marginal costs, barely any actual revenue is generated.

The traditional massive open online courses (or MOOC) essentially belong to the online publishing industry. From Yang's perspective, MOOC founder Coursera is simply an online publishing company and the product of a manufacturing mindset: just as one develops a textbook and hopes for hundreds of people to read it. Teachers develop courses for students to purchase, but with little to no interaction between teacher and student.

However, the awareness of users willing to pay for services is growing. Yang makes a point of including education as a part of the service industry, maintaining that both online and offline teachers deserve payment so long as students recognize they have utilized the teachers' time.

Is AI technology the solution for "diseconomies of scale?"

Everybody is a differentiated individual who desires a personalized service. But whether a personalized study platform can realize expansion in scale has become the key to the success of the business model.

The education industry has been facing the structural contradictions of "diseconomies of scale." With the mass expansion in student numbers, companies' costs on marketing, management and teachers have increased, reflected in the losses of the companies referred to above.

The gross profit margin of foreign teachers ranges from about 40% to 60%. 51Talk, an online English education platform that has been listed on the New York Stock Exchange, post a gross profit margin of 64.6% in its latest quarterly financial report. Meanwhile, Dada founder and CEO Zhi Hui has revealed their platform has a gross profit margin of around 60%.

iTutorGroup entered the 1-to-1 foreign teacher learning market early on. Bloomberg reported the company has already realized an 82% gross profit margin, well above the industry average.

Eric Yang - founder, chairman and CEO of iTutorGroup

Yang has confirmed the above-mentioned statistic and attributes the high gross profit margin to the technology tools they have developed with AI and big data, most significantly DCGS (Dynamic Course Generation System).

DCGS consists of data collection and algorithms. In terms of data collection, DCGS collects big data based on students' demographic characteristics and real-time class attendance, including age, gender, occupation, location, learning habits, hobbies and behavior across multiple devices, as well as their interactions with foreign teachers and schoolmates.

As for algorithms, DCGS analyzes collected data through optimization algorithms, making dynamic adjustments to match students with the most suitable foreign teachers, classmates and learning content. After the class, DCGS reviews the student feedback of foreign teachers, teaching materials and classmates to further understand user demands and preferences and constantly improve the back-end database.

Online education companies are willing to embrace new technologies. At present, AI in the education industry is applied through facial recognition technology, real-time capture of changes in performance, and accumulation of personal data. It is then able to analyze each person's strengths and weaknesses before recommending them the most suitable learning services, thus improving education experience and efficiency.

Improving the standardization of the teaching process through AI has to some extent reduced difficulties for new teachers while breaking the industry's teacher supply bottleneck.

Yang believes that human facial and expression recognition are the most obvious applications of AI in the education industry, and can generate real value at the operational maintenance and sales management levels.

For instance, the system assigns 128 tags to each student, which are regarded as the students' DNA, and based on which the platform implements coping strategies. For instance, the system can learn the probability of a student’s demand for refund on the first day after they pay the tuition.

The logic behind the algorithms is this: if a student finishes 16 lessons on the platform within one month, it can be concluded the student's renewal rate has exceeded 90%. Thus one of the most important tasks of the platform's learning manager assistant is to confirm if the users have taken 16 hours of classes or not.

If the system finds a student’s tags coincide with the characteristics of those more likely to apply for a refund, the platform recommends 1-on-1 service and selects a teacher matching the student's preference. If the 1-to-1 lessons go ahead problem-free, the platform will then recommend 2-to-1 or 3-to-1 sessions, thus realizing higher gross profits through better economies of scale.

In contrast to the monitoring of facial expressions, the amount of speaking time is of vital importance in collecting student data, with the degree of continuity and pause in open speech being a basis for judging students’ fluency. Active companions are arranged for shy students to complete the class. The collection of micro-data coupled with algorithms is utilized to achieve a high level of individualization in the online classroom.

“The operation mode should be refined in order to achieve ideal results. The age of students or their English level is one dimension; the school's curriculum can be another. For adult students, there’ll also be a breakdown of the industry in which he/she works. This is the refined operational mindset.”

Teacher management is also a top priority. iTutorGroup now has about 20,000 foreign teachers from over 80 countries. Cross-time zone and cross-regional management is another important challenge for the platform.

Yang also recognizes teachers as users on the platform. He says the core of teacher management lies in whether they can achieve a reasonable amount of class hours. The more class hours teachers get, the more they depend on the platform, leading to a higher level of stability in the teaching team.

Algorithms plays a role in enabling teachers to acquire more class hours, thus earning more income from the platform. The more positive reviews a teacher receives, the more students the system recommends to the teacher. Once a teacher's class hours accumulate to a certain degree, his/her level goes up. Such incentive mechanisms are built into the system.

If a teacher leaves the platform, all of his/her accumulated reviews must be reset in order to avoid other platforms tapping up the teachers and offering more premium salaries.

iTutorGroup's investment in technology has been increasing by approximately 40% year on year. AI is utilized in almost all departments, from learning to teaching, customer service and sales management. Through big data collection and algorithm matching, AI can meet the personalized needs of its user base. According to Yang, this not merely serves as the basis for iTutorGroup to successfully carry out its one-to-many teaching model and increase its gross profit margin, but also represents the major weapon and form of future companies to come.

--------------------------------------------------

Follow us now on Twitter @tmtpostenglish, Medium @TMTpost and Facebook @tmtpost)

The Chinese version article is published with authorization from the author @Li Chengcheng. The English version article is edited by @Yi LIU. Please note source and hyperlink when reproduce.