Differentiated Competitive Strategy and Industry-Leading Path of

Online Education Platform: Take ZuoYeBang as an Example

Jiaxuan Yu

, Dian Zhang

and Dingyao Zhang

School of Economics and Management, Beijing Jiaotong University, Shangyuan Village, Haidian District, China

Department of E-commerce School of Tourism, Jinan University, Huangpu Avenue, Tianhe District, China

Glorious Sun School of Business & Management, Donghua University, Zhongshan West Road, Changning District, China

Keywords: ZuoYeBang, Competitive Differentiation, Large AI Model, Sinking Market, Ecological Closed Loop.

Abstract: In recent years, China's online education industry has ushered in rapid development driven by policy support

and technological innovation, but the market competition in the industry has also become increasingly fierce.

This paper takes the company as the research object and systematically discusses how it achieves industry

leadership through its unique strategy, as well as the construction path and implementation effect of its

differentiated competitive strategy through case analysis, comparative research and quantitative data analysis.

The study found that through the three core strategies of technology tooling, sinking market penetration and

closed-loop software and hardware ecology, ZuoYeBang has successfully created an exclusive competitive

advantage, which in turn promoted the simultaneous increase of user base and business value, and achieved

continuous growth in market share. At the same time, in view of potential risks such as policy adjustment,

acceleration of technology iteration and low-price competition in the market, this paper also analyzes the

coping strategies such as content transformation, data compliance upgrade and supply chain diversification,

which provides a replicable strategic framework for EdTech enterprises to respond to policy supervision and

optimize technology paths, and provides a theoretical reference for the sustainable development of the

industry.

1 INTRODUCTION

China's online education industry is under the triple

drive of policy guidance, technological innovation

and surging market demand, and has ushered in new

development opportunities. At the same time, with

the continuous advocacy of the country's concepts of

educational fairness and quality education, policy

supervision has become stricter, and industry

competition has become more and more intense. In

this context, major online education platforms have

explored differentiated development paths in order to

occupy an advantage in market competition. As an

industry leader, ZuoYeBang has risen rapidly and

achieved remarkable results by virtue of its unique

competitive strategy and technical advantages.

According to statistics, in 2024, the market share of

the online learning machine under ZuoYeBang will

https://orcid.org/0009-0005-0787-2180

https://orcid.org/0009-0006-6505-356X

https://orcid.org/0009-0006-7019-1588

be as high as 33%, and its products have covered 90%

of the prefecture-level cities in the country, which not

only reflects the strong penetration of the job gang in

the market but also reflects its precise regional

positioning and product strategy.

The purpose of this study is to explore how

ZuoYeBang can achieve a leading position in the

education technology industry through differentiated

competitive strategies, so as to provide a useful

reference for the transformation and upgrading of

online education platforms. The research focuses on

three core dimensions: technology application,

market positioning, and ecological synergy. In terms

of technology, relying on the self-developed "Galaxy

Model" and DeepSeek-R1 inference model, and using

cutting-edge technologies such as OCR, speech

recognition and natural language processing,

ZuoYeBang has realized full-discipline correction

460

Yu, J., Zhang, D. and Zhang, D.

Differentiated Competitive Strategy and Industry-Leading Path of Online Education Platform: Take ZuoYeBang as an Example.

DOI: 10.5220/0013847100004719

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 2nd International Conference on E-commerce and Modern Logistics (ICEML 2025), pages 460-464

ISBN: 978-989-758-775-7

and personalized weakness diagnosis, greatly

improved the response speed of problem solving, and

continuously optimized the AI model through data

feedback to provide users with accurate and efficient

learning assistance (Chen, 2022). In terms of the

market, by aiming at the sinking market, ZuoYeBang

has formulated a low-price strategy and a regional

adaptation plan, and launched learning hardware

products that meet the local consumption level,

forming a wide range of market coverage and user

groups. In terms of ecological collaboration,

ZuoYeBang has built a closed-loop operation model

of "tools, content, and communities", and through

multi-channel linkage such as online apps, live

courses, and offline agents, it not only realizes the

exchange of hardware and software data, but also

enhances user stickiness and brand loyalty.

In addition, in order to more comprehensively

analyze the unique advantages of Job Bang in the

fierce market competition, this study also compares

the strategic choices of ZuoYeBang's main

competitors. Different from the practice of some

competitors focusing on live courses and

standardized teaching content, ZuoYeBang pays

more attention to the practice of technical tools, and

naturally guides users from "searching for questions"

to a complete closed loop of "learning-practice-

testing" through functions such as photo search. This

development model supported by technology, with

market segmentation as the starting point, and

ecological construction as the path, has not only won

the reputation of users for the ZuoYeBang, but also

provided new ideas for the transformation and

upgrading of the entire industry (Liu et al., 2022).

In summary, this study aims to explore its

successful experience in technological innovation,

market segmentation, and ecological construction by

integrating the industry background, policy

environment, and specific practices of the job gang,

and further analyze its response strategies in the

future to cope with the multiple challenges of policy,

technology, and market. This research not only helps

to reveal the development of online education

platforms in the new era, but also provides valuable

theoretical and practical references for other

education enterprises.

2 DIFFERENTIATION

STRATEGY ANALYSIS

In the fierce competition of online education,

ZuoYeBang has built a differentiated competitive

advantage that is difficult to replicate with its unique

technological innovation, market positioning and

ecological synergy model. From the analysis of

technology, market, ecology and service model, it can

be seen that its strategic layout has formed effective

competitive barriers in all links, ensuring the

continuous improvement of user experience and

brand loyalty.

First of all, the technical differentiation of

ZuoYeBang is reflected in the in-depth application

and instrumental innovation of AI large models. After

the integration of its self-developed "Galaxy Large

Model" and DeepSeek-R1 inference model, it has

achieved a breakthrough of 99.9% accuracy and

problem-solving response speed within 1 second,

which is significantly better than the industry

average. According to the data of the C-Eval

rankings, the Galaxy model leads the 2023 evaluation

with a comprehensive score of 73.7, especially in the

social sciences (86 points) and humanities (71.6

points), far surpassing Mengzi (71.5 points) and GPT-

4 (68.7 points), as shown in Table 1. This technical

advantage is directly translated into the improvement

of user experience, and the average daily call volume

of its OCR photo search function has exceeded 30

million times, and users have extended from a single

search behavior to learning diagnosis and

consolidation of wrong questions, forming a closed-

loop of the whole process of "learning-practice-

testing", and building a high-frequency use scenario

with an average of 68 minutes per day (Ma and

Anekiti, 2024).

At the same time, its technology patent reserve

also further strengthens its advantages. ZuoYeBang

has more than 200 patents in the fields of OCR,

speech recognition and natural language processing,

and its OCR response speed is 30% faster than that of

competing products. The data exchange between

hardware and software forms a positive feedback

mechanism, and the average daily usage time of

learning machine users is 68 minutes, which is much

higher than the data performance of competitors in

about 45 minutes (Aurora Mobile, 2021).

Empowered by technological advantages,

homework help's market positioning strategy presents

dual characteristics of precise focus and scene

extension. Taking photos and searching questions can

meet the demand of instant answering questions. The

live class and the double-teacher class can solve the

pain points of systematic learning, while the

intelligent question bank and the wrong question

book function are deeply bound to the after-school

review scene, and the scale is expanded through

technology-driven low-cost operation (Huang, 2024).

Differentiated Competitive Strategy and Industry-Leading Path of Online Education Platform: Take ZuoYeBang as an Example

461

Table1：C-Eval Global Model Comprehensive Exam Evaluation Ranking Table

Name Issuing Authority Submission

time

Average Average

(Hard)

STEM social

science

humanities other

Galaxy Zuoyebang 2023/8/23 73.7 60.5 71.4 86 71.6 68.8

Mengzi LangBoat 2023/8/25 71.5 48.8 62.3 87.2 76.8 68.6

ChatGLM2 Tsinghua&Zhipu.AI 2023/6/25 71.1 50 64.4 81.6 73.7 71.3

UniGPT2.0 Unisound 2023/8/28 70 52.8 65.7 78.7 67 72.9

360GPT-S2 360 2023/8/29 69 42 59.4 82 70.6 72.9

InternLM-

123B

Shanghai AI Lab &

SenseTime

2023/8/22 68.8 50 63.5 81.4 72.7 63

GPT-4 Open AI 2023/5/15 68.7 54.9 67.1 77.6 64.5 67.8

AiLMe-

100B v2

APUS 2023/7/25 67.7 55.3 65.4 72.3 71.2 64

Source: C-Eval Leaderboard

In terms of the market, ZuoYeBang achieves

differentiated competition by precisely targeting the

lower-tier market. Its learning machines are priced

starting from 1,999 yuan, significantly lower than

those of iFLYTEK (over 5,000 yuan), which aligns

with the purchasing power of cities below the third

tier. Data from 2024 shows that ZuoYeBang's sales in

the lower-tier market account for 70%, with localized

textbooks adapted to 260 versions, covering 95% of

the country's regions. In contrast, TAL focuses on

high-end courses in first- and second-tier cities (with

a user growth rate of 18%), while Yuanfudao mainly

offers online live-streaming courses (covering 80% of

first- and second-tier cities). ZuoYeBang, however,

holds an advantage with a 45% user growth rate in the

lower-tier market and a 32% hardware penetration

rate (compared to 10% for TAL and 8% for

Yuanfudao) (Mobile Education Platform Zuoyebang,

2020).

Furthermore, ZuoYeBang has further

consolidated its market position through a full-

channel coverage strategy. It leads in sales in the

2,000-3,000-yuan price range on JD.com and Tmall.

Offline, it covers 90% of prefecture-level cities

through agents, forming a conversion chain of "low-

price attractions - hardware binding - service value-

added". Its user segmentation operation strategy has

also been highly effective: the P series (cost-

effective), T series (mid-range), and X series (high-

end) product matrix meets different needs, with a paid

user conversion rate of 32% (industry average is

25%), and a year-on-year increase of 15% in

repurchase rate (Pan and Mao, 2025).

In terms of ecological collaboration and service

models, ZuoYeBang has established a closed-loop

ecosystem cantered on "tools + content +

community". Besides its core learning machine

product, ZuoYeBang has successively launched a

variety of educational hardware such as smart

learning desks, error printers, and word cards,

forming a complete product matrix from basic to

high-end models, covering the needs of different

users from preschool to high school. Relying on the

Zuoyebang APP, which has a monthly active user

base of 170 million, the platform has realized

functions such as photo search for questions, video

Q&A, and parent-end learning situation reports.

Through data intercommunication between hardware

and the APP, a feedback mechanism from user

behavior to product optimization has been

established. Data shows that 70% of learning machine

users also use the APP's question bank. This cross-

scenario product interaction not only enhances user

stickiness but also forms a positive data feedback

loop, further optimizing the AI model. In terms of

community-based operation, ZuoYeBang has created

a highly engaged user community through methods

such as parent-end point redemption, real-time

learning situation monitoring, and online live

interaction (Liu et al., 2022). According to relevant

survey reports, the parent-end participation rate of

ZuoYeBang is as high as 80%, far exceeding the

average level of competitors. This community-based

operation model effectively enhances brand loyalty

and user activity. The community-based operation

strategy further strengthens the ecological closed

loop. Additionally, the "hardware + lifetime

membership" bundle (with a conversion rate of 28%)

and the "trade-in" service (with a repurchase rate of

ICEML 2025 - International Conference on E-commerce and Modern Logistics

462

35%) have established a long-term retention

mechanism, forming a competitive barrier (Gu,2024).

Overall, ZuoYeBang has formed a distinct

competitive advantage through precise layout in the

three major areas of technology, market, and

ecological services. Its technology toolization

strategy not only improves teaching efficiency but

also builds a solid data barrier; while its low-price

strategy-centered layout in the lower-tier market

successfully breaks through the geographical

limitations of traditional educational resources; at the

same time, the "tools + content + community" closed-

loop ecosystem brings lasting user stickiness to the

brand. It is precisely this multi-dimensional

coordinated development model that keeps

ZuoYeBang at the forefront of the highly competitive

online education market, providing a referenceable

development path and valuable experience for other

educational platforms.

3 POTENTIAL CHALLENGES

AND RESPONSE STRATEGIES

As a leading enterprise in China's education

technology sector, ZuoYeBang has performed

outstandingly in the smart hardware market and with

its differentiated corporate strategies. However, the

potential challenges it faces are becoming

increasingly prominent. Firstly, it is confronted with

policy risks from the state. One is the deepening of

the "Double Reduction" policy, where the Ministry of

Education further restricts "hidden subject training"

in 2024, requiring that educational hardware content

must not include subject-based problem-solving

videos, which directly affects the iteration of

ZuoYeBang's "AI Question Bank" function (where

the proportion of subject-based content in the original

function was 65%). The second is data security

reviews. The Personal Information Protection Law

stipulates that educational enterprises must pass

"security assessment and certification" when storing

user data. According to relevant data from the

Cyberspace Administration, ZuoYeBang was

summoned twice in 2024 due to cross-border data

transmission issues. To address these policy risks,

ZuoYeBang can take corresponding measures. Firstly,

it can transform its course content. For instance, it

could launch non-subject-based courses such as

"Science Experiment Classes" and "Programming

Thinking Classes" in 2024, increasing their

proportion from 15% to 40%. Secondly, it can

upgrade its data compliance. For example, it could

invest 120 million yuan in building a localized data

center and obtain ISO 27001 certification, ensuring a

100% compliance rate for user data storage (The Fly,

2021).

Secondly, ZuoYeBang also faces technical risks.

One is the risk of outdated AI models. The average

iteration cycle of large-scale educational models has

been shortened to six months (with an average

industry R&D cost of 50 million yuan per cycle).

According to industry reports, the comprehensive

score of Zuoyebang's Galaxy Large Model dropped

from 73.7 in 2023 to 70.2 in 2024. The other is the

fluctuation in the hardware supply chain. In Q2 of

2024, the main control chip for learning machines

(MediaTek MT8696) had a 30-day delay in delivery

due to insufficient capacity, resulting in a direct loss

of 300 million yuan in sales. To address these

technical risks, ZuoYeBang can take corresponding

measures. Firstly, it can engage in technical alliance

cooperation, such as jointly establishing the

"Intelligent Education Joint Laboratory" with

Tsinghua University and the Institute of Automation

of the Chinese Academy of Sciences. In 2024, they

jointly published 23 AI patents, reducing the single-

point R&D cost. Secondly, it can diversify the supply

chain, such as introducing Rockchip and

Unisplendour as backup chip suppliers. In 2024, the

procurement cost decreased by 18% (Zhang et al.,

2021).

Finally, ZuoYeBang also faces market risks. The

first is the impact of low-price competition.

ByteDance's "Dali Education" launched a learning

machine priced at 999 yuan (with a configuration

comparable to ZuoYeBang's P series), causing the

user churn rate in the lower-tier market of

ZuoYeBang to rise from 8% to 15%. The second is

the decline in users' willingness to pay: in 2024, the

average budget for educational hardware per K12

parent dropped from 2,500 yuan to 1,800 yuan, and

the return rate of ZuoYeBang's learning machines

increased by 5% year-on-year. To address these

market risks, ZuoYeBang can take corresponding

measures. Firstly, it can introduce differentiated

value-added services, such as launching a "hardware

+ lifetime membership" bundle (priced at 2,499 yuan,

including 1,200-yuan worth of quality courses),

which increased the conversion rate to 28% in 2024

(industry average 12%). Secondly, it can implement

user retention plans, such as offering "trade-in + data

migration" services for existing users, raising the

repurchase rate from 20% to 35%.

Differentiated Competitive Strategy and Industry-Leading Path of Online Education Platform: Take ZuoYeBang as an Example

463

4 CONCLUSION

This article employs the case analysis method to

examine how ZuoYeBang has achieved industry

leadership through a differentiation strategy. It

concludes that ZuoYeBang's leading position stems

from the synergistic effect of three core strategies.

The first is the toolization of technology,

transforming AI into practical educational tools (such

as photo search for questions and weak point

diagnosis), creating high-frequency usage scenarios,

and building user habits and data barriers. The second

is market niche competition: by adopting a low-price

strategy and regional adaptation, it has captured the

lower-tier market, avoiding direct competition with

Yundaixue and TAL, and opening up new growth

space. The third is the positive feedback of the

ecosystem: the data loop between hardware and

software, as well as the reuse of technology

domestically and internationally, form a flywheel

effect from user growth to data accumulation and then

to experience optimization. Furthermore, it elaborates

on the potential challenges faced by ZuoYeBang,

specifically addressing the corresponding

countermeasures for the three major risks of policy,

technology, and market.

As a leading enterprise in the education

technology industry, ZuoYeBang has achieved

remarkable results through its differentiated strategy

and taken the lead in the industry. For its future

development, it needs to balance opportunities and

challenges in multiple dimensions and continuously

optimize its differentiated strategy. It is expected to

continue to lead in the trillion-yuan-level technology

education market and move towards a global

education technology platform.

AUTHORS CONTRIBUTION

All the authors contributed equally and their names

were listed in alphabetical order.

REFERENCES

Aurora Mobile, 2021. Aurora Mobile partners Zuoyebang

to promote quality growth of its online education app.

Benzinga Newswires. https://www.proquest.com/wire-

feeds/aurora-mobile-partners-zuoyebang-promote-

quality/docview/2518313411/se-2.

Chen, D., 2022. Application of IoT-oriented online

education platform in English teaching. Mathematical

Problems in Engineering 2022.

Gu, Z. Research on the accuracy of machine learning-based

AI grading systems in handling high school math

function problems: A comparative study of

MathGPTPro and Zuoyebang, 2024. Lecture Notes in

Education Psychology and Public Media.

Huang, Y., 2024. Investigation on the current situation of

the use of practice-assistance APP in high school

physics learning and research on influencing factors.

Southwest University.

Liu, Y. et al., 2022. Online homework intelligent platform

based on self-regulated learning (SRL): Essential for

sustainable development of online higher education.

Sustainability 14(24), 16904.

Ma, Y., Anekiti, W., 2024. Comparative study on

intelligent learning tools for mathematics in primary

and secondary schools. Primary and Secondary School

Electrification Education (12), 14-19.

Mobile Education Platform Zuoyebang Wins US$ 750

Million Series E, 2020. China Telecom 27(6), 8.

Pan, S., Mao, Y., 2025. Using AI to "cheat" and making

Apps "take off". IT Times 006.

The Fly, 2021. Aurora Mobile enters partnership agreement

with Zuoyebang Education. Gale General OneFile.

https://link-gale-

com.ezproxy.rit.edu/apps/doc/A656247832/ITOF.

Zhang, J. et al., 2021. A data-driven analysis of K-12 stude

nts’ participation and learning performance on an onlin

e supplementary learning platform. In 2021 Internation

al Conference on Computer Communications and Net

works (ICCCN) (pp. 1-9). Athens, Greece.

ICEML 2025 - International Conference on E-commerce and Modern Logistics

464