Can China’s Civil-Service Exam Training Market Build Its Own “Beike” Platform Model?
A price war in China's civil-service exam training market broke into the open in September, when Fenbi (HKEX: 2469) publicly broke down a competitor's cost structure and launched a product closely resembling Chaoke Education's National Day boot camp at RMB 1,380 against the rival's RMB 2,180. It sold more than 1,200 units on day one.
The market itself is still growing. Spending on recruitment-exam training is forecast to rise from RMB 30.5 billion in 2021 to RMB 50.5 billion in 2026, and the number of institutions had passed 15,000 by the end of January 2026, more than 8,000 of them small and mid-sized. Candidates passing review for the national civil-service exam grew from 1.576 million in 2021 to a record 3.718 million in 2026, against 38,100 openings. The share paying for coaching is expected to exceed 31% this year, up from about 27% in 2021.
Scale has not translated into comfortable margins. Wang Qingying, who spent years at Zhonggong Education before joining Chaoke Education in May, argues the core problem is no repeat purchase: passing means the customer leaves for good, and failing usually means switching schools.
The arithmetic is unforgiving. For a mid-sized offline operator with around 500 staff, Wang estimates payroll alone can exceed RMB 5 million a month and total outlay can top RMB 10 million, so gross billing must reach at least RMB 13 million a month to break even and absorb refunds from agreement classes, which return tuition if the candidate does not pass.
That pressure feeds the price war. A day of written-exam coaching that once commanded RMB 1,000 has fallen to as little as RMB 100 in some markets. At the other end, offline interview courses still run at RMB 1,800 to RMB 2,000 a day and packages reach RMB 50,800 for two years of unlimited study.
Organisation is changing too. Large providers have been cutting salaried teachers and rehiring them as freelancers paid by the hour, shifting risk onto individuals; some sales teams now work without a base salary, organised as partners splitting 50% of profit. Agencies broker teacher supply, but in peak season they may plug in under-prepared instructors, and teachers serving several institutions at once struggle to keep up with provincial exam differences, what Wang calls one set of lecture notes for the whole country.
Two operating models are diverging. One keeps building prefecture-level branches and residential training bases, with high fixed costs but direct local reach. The other acquires students through online traffic, concentrates teaching in a provincial capital and keeps only light local service staff, trading fixed cost for a heavier dependence on content and conversion.
Franchising, once the standard route for new entrants, is losing appeal. Wang advises against it: a head office cannot dispatch its star teachers to every franchise during peak season, quality is hard to guarantee locally, and renewal terms can change once the franchisee has built the market. He estimates entry costs start at around RMB 2 million for a prefecture-level operation and RMB 300,000 to RMB 500,000 for a county branch, with payback taking four to five years.
Hence the Beike question. KE Holdings built an agent cooperation network that lets brokers from different brands complete a transaction together and split proceeds by contribution. Exam training has shareable assets - enrolment channels, teachers, curriculum R&D - but teaching resists the standardisation that made housing listings tradable, and revenue split, quality assurance and accountability remain unresolved. His forecast: most small and mid-sized players exit within five years, leaving national brands and small specialists built around a strong teacher IP.
The market itself is still growing. Spending on recruitment-exam training is forecast to rise from RMB 30.5 billion in 2021 to RMB 50.5 billion in 2026, and the number of institutions had passed 15,000 by the end of January 2026, more than 8,000 of them small and mid-sized. Candidates passing review for the national civil-service exam grew from 1.576 million in 2021 to a record 3.718 million in 2026, against 38,100 openings. The share paying for coaching is expected to exceed 31% this year, up from about 27% in 2021.
Scale has not translated into comfortable margins. Wang Qingying, who spent years at Zhonggong Education before joining Chaoke Education in May, argues the core problem is no repeat purchase: passing means the customer leaves for good, and failing usually means switching schools.
The arithmetic is unforgiving. For a mid-sized offline operator with around 500 staff, Wang estimates payroll alone can exceed RMB 5 million a month and total outlay can top RMB 10 million, so gross billing must reach at least RMB 13 million a month to break even and absorb refunds from agreement classes, which return tuition if the candidate does not pass.
That pressure feeds the price war. A day of written-exam coaching that once commanded RMB 1,000 has fallen to as little as RMB 100 in some markets. At the other end, offline interview courses still run at RMB 1,800 to RMB 2,000 a day and packages reach RMB 50,800 for two years of unlimited study.
Organisation is changing too. Large providers have been cutting salaried teachers and rehiring them as freelancers paid by the hour, shifting risk onto individuals; some sales teams now work without a base salary, organised as partners splitting 50% of profit. Agencies broker teacher supply, but in peak season they may plug in under-prepared instructors, and teachers serving several institutions at once struggle to keep up with provincial exam differences, what Wang calls one set of lecture notes for the whole country.
Two operating models are diverging. One keeps building prefecture-level branches and residential training bases, with high fixed costs but direct local reach. The other acquires students through online traffic, concentrates teaching in a provincial capital and keeps only light local service staff, trading fixed cost for a heavier dependence on content and conversion.
Franchising, once the standard route for new entrants, is losing appeal. Wang advises against it: a head office cannot dispatch its star teachers to every franchise during peak season, quality is hard to guarantee locally, and renewal terms can change once the franchisee has built the market. He estimates entry costs start at around RMB 2 million for a prefecture-level operation and RMB 300,000 to RMB 500,000 for a county branch, with payback taking four to five years.
Hence the Beike question. KE Holdings built an agent cooperation network that lets brokers from different brands complete a transaction together and split proceeds by contribution. Exam training has shareable assets - enrolment channels, teachers, curriculum R&D - but teaching resists the standardisation that made housing listings tradable, and revenue split, quality assurance and accountability remain unresolved. His forecast: most small and mid-sized players exit within five years, leaving national brands and small specialists built around a strong teacher IP.
JMDedu is a leading B2B EdTech Media in China that informs and connects business professionals through our reports, events, and information services. Our mission is to drive the advancement of education. Our network spans the education industry in China as well as globally, including entrepreneurs, investors, startups, educators, business decision-makers, and policy decision-makers.
Copyright © 2014-2026 en.jmdedu.com All Rights Reserved