Survey of China's Small Education Entrepreneurs: 'Just Getting By' Is No Longer Enough
Jiemodui surveyed its community of education entrepreneurs in October 2026, inviting more than 300 founders and collecting 78 valid responses. The resulting report describes a sector in which revenue, cash runway and stated confidence increasingly point in different directions. One respondent in Qingdao, who has run a nine-year-old study-tour business, chose a single phrase as his keyword for 2026: "just getting by."
Official data explains part of the pressure. Drawing on five rounds of the CIEFR-HS household survey between 2017 and 2025, Peking University researchers found participation in academic tutoring among primary and secondary students fell from 38.1% to 20.1%, and in non-academic tutoring from 22.4% to 16.7%. Average per-student spending nonetheless dropped only 11.8%, to RMB 2,873 from RMB 3,257, because those still paying pay more: in 2025 the highest-spending 10% of students accounted for 75.8% of all out-of-school tutoring expenditure. On the supply side, the Ministry of Education has said offline academic tutoring institutions for compulsory education were cut from 124,000 to 9,728, a reduction of 92.14%. In the first half of 2026 the education sector recorded just 16 financing deals with about RMB 121 million disclosed — both six-year lows for the period.
The founders surveyed are not short of demand, but the demand that converts is narrow. A 29-year-old STEM education founder in Shanghai, running a two-year-old business with two full-time and seven part-time staff and about RMB 500,000 in annual revenue, said customers pay for outcomes — competition awards — while he is teaching observation, reasoning and problem-solving. A four-year-old high-school tutoring provider in Lu'an, Anhui, with 20 full-time and 15 part-time staff and roughly RMB 8 million in revenue, lifted prices about 15% and still grew revenue around 20% in 2026. A ten-year-old Shanghai STEAM provider with about RMB 800,000 in revenue is cutting consumer prices and raising school-side prices, and its founder calls herself pessimistic about the industry.
The second theme is where founders spend their time. Content production, channel maintenance, product development and front-line delivery are converging on one person. In Fuzhou, a civil-service exam training founder saw revenue fall about 30% to RMB 1.5 million and now records video and works on traffic conversion. A one-year-old Beijing AI founder community, with RMB 3 million in revenue, grew 80% through podcasts and video interviews. But a five-year-old education digitisation firm in Xi'an earns under RMB 200,000 a year despite heavy publishing. Citing Oliver Williamson's transaction-cost economics, the report argues that AI dissolves only part of the transaction costs in standardised work, while opportunism risk persists in high-value collaboration and asset specificity remains the binding constraint on very small operators.
Pivots look less like choosing a future than betting on adjacent capability — the report invokes Stuart Kauffman's "adjacent possible" and James March's exploration-versus-exploitation tension. A seven-year-old AI learning community in Foshan, once above RMB 5 million in revenue, saw revenue drop 40% and cut its events and video teams to sell AI services and training to pharmaceutical clients. An 11-year-old study-tour agency in Ningbo with more than RMB 8 million in revenue is repositioning around travel for older customers, using membership deposits and premium pricing.
The sharpest point concerns language. A two-year-old Hangzhou AI bootcamp reports roughly three months of cash left while its founder calls this the best era for entrepreneurs; a nine-year-old Beijing science education company with 55 full-time staff booked a 40% revenue decline, yet remains convinced that "AI plus" is the trend. Revenue, cash positions and actions do not move together, and optimism is not by itself evidence about operating judgement. "The future can be imagined," the report concludes. "The present has to be accounted for."
Official data explains part of the pressure. Drawing on five rounds of the CIEFR-HS household survey between 2017 and 2025, Peking University researchers found participation in academic tutoring among primary and secondary students fell from 38.1% to 20.1%, and in non-academic tutoring from 22.4% to 16.7%. Average per-student spending nonetheless dropped only 11.8%, to RMB 2,873 from RMB 3,257, because those still paying pay more: in 2025 the highest-spending 10% of students accounted for 75.8% of all out-of-school tutoring expenditure. On the supply side, the Ministry of Education has said offline academic tutoring institutions for compulsory education were cut from 124,000 to 9,728, a reduction of 92.14%. In the first half of 2026 the education sector recorded just 16 financing deals with about RMB 121 million disclosed — both six-year lows for the period.
The founders surveyed are not short of demand, but the demand that converts is narrow. A 29-year-old STEM education founder in Shanghai, running a two-year-old business with two full-time and seven part-time staff and about RMB 500,000 in annual revenue, said customers pay for outcomes — competition awards — while he is teaching observation, reasoning and problem-solving. A four-year-old high-school tutoring provider in Lu'an, Anhui, with 20 full-time and 15 part-time staff and roughly RMB 8 million in revenue, lifted prices about 15% and still grew revenue around 20% in 2026. A ten-year-old Shanghai STEAM provider with about RMB 800,000 in revenue is cutting consumer prices and raising school-side prices, and its founder calls herself pessimistic about the industry.
The second theme is where founders spend their time. Content production, channel maintenance, product development and front-line delivery are converging on one person. In Fuzhou, a civil-service exam training founder saw revenue fall about 30% to RMB 1.5 million and now records video and works on traffic conversion. A one-year-old Beijing AI founder community, with RMB 3 million in revenue, grew 80% through podcasts and video interviews. But a five-year-old education digitisation firm in Xi'an earns under RMB 200,000 a year despite heavy publishing. Citing Oliver Williamson's transaction-cost economics, the report argues that AI dissolves only part of the transaction costs in standardised work, while opportunism risk persists in high-value collaboration and asset specificity remains the binding constraint on very small operators.
Pivots look less like choosing a future than betting on adjacent capability — the report invokes Stuart Kauffman's "adjacent possible" and James March's exploration-versus-exploitation tension. A seven-year-old AI learning community in Foshan, once above RMB 5 million in revenue, saw revenue drop 40% and cut its events and video teams to sell AI services and training to pharmaceutical clients. An 11-year-old study-tour agency in Ningbo with more than RMB 8 million in revenue is repositioning around travel for older customers, using membership deposits and premium pricing.
The sharpest point concerns language. A two-year-old Hangzhou AI bootcamp reports roughly three months of cash left while its founder calls this the best era for entrepreneurs; a nine-year-old Beijing science education company with 55 full-time staff booked a 40% revenue decline, yet remains convinced that "AI plus" is the trend. Revenue, cash positions and actions do not move together, and optimism is not by itself evidence about operating judgement. "The future can be imagined," the report concludes. "The present has to be accounted for."
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