8 million users say goodbye to Keep

Ask AI · Can the model of Keep, which relies on cost reduction for profitability but is losing users, be sustainable?

The leading player in the internet fitness sector, Keep, has delivered a contradictory performance report for 2025: annual operating revenue fell by 20.7% year-on-year, over 8 million monthly active users were lost, and all three core revenue segments declined, yet it achieved adjusted net profit for the first time in its history.

China’s Jin Tong News noted that this accounting profit, achieved through extreme cost-cutting, is not driven by positive growth in its core business; rather, it accompanies a significant shrinkage in user scale and a historic low in stock price. The capital market has voted with its feet, as its stock price in Hong Kong has plummeted by 90% from its issuance price, while the AllinAI transformation strategy is still in the testing phase, lacking actual growth momentum.

This “thrift” winning over “expansion” for Keep’s profitability raises the question of whether it is merely a temporary financial optimization or a desperate measure, becoming a focal point of industry concern.

Core Business in Decline

On March 25, Keep released its 2025 annual performance announcement, showing that the company achieved operating revenue of 1.637 billion yuan, a year-on-year decrease of 20.7%; gross profit was 854 million yuan, down 11.5%, but adjusted net profit reached 25.216 million yuan, marking the first annual adjusted profit in the company’s development history.

However, behind this profit lies the reality of a comprehensive decline in core business, as all three revenue segments fell victim to the downward trend, and profitability did not stem from “increased revenue” but rather from extreme “cost reduction.”

From the revenue structure perspective, Keep’s own brand sports product revenue was 778 million yuan, a year-on-year decline of 18.4%; online membership and paid content revenue was 680 million yuan, a significant year-on-year drop of 25.9%, making it the segment with the largest decline, directly reflecting the continued weakening of user willingness to pay; advertising and other income was 179 million yuan, down 7.5% year-on-year.

Regarding the revenue decline, Keep attributed it to the proactive reduction of low-margin hardware products like home fitness equipment and wearables, as well as the fading popularity of blockbuster IP events in 2024 leading to a significant shrinkage in online sports event revenue, but this cannot hide the overall downturn in its business development.

In the context of a comprehensive decline in core business, Keep’s profitability is entirely reliant on stringent cost control and contraction.

Data shows that in 2025, Keep’s operating costs were 783 million yuan, a significant decrease of 28.8% compared to 2024, with the intensity of cost control far exceeding the revenue decline. Among these, the reduction in marketing expenses was the most significant, with sales and marketing expenditure dropping from 758 million yuan in 2024 to 439 million yuan, a year-on-year decrease of 42.1%, and spending on brand and market promotion alone reduced by 283 million yuan, with customer acquisition investment nearly “halved.”

R&D expenditures also pressed the contraction button, with R&D spending of 310 million yuan down 29.4% from 439 million yuan in 2024, and personnel costs for R&D and outsourcing reduced by 85.3 million yuan and 24.1 million yuan, respectively.

Personnel optimization became the core focus of this cost reduction, with Keep ending 2025 with only 645 full-time employees, down 182 from 827 at the end of 2024, a year-on-year decrease of 22%. Looking at a longer time frame, the reduction in personnel is even more striking, as the company had 1,243 full-time employees at the end of 2022, a reduction of nearly 600 over three years, with a decrease of 48.11%, and employee benefits expenditure also fell from 648 million yuan to 515 million yuan.

Although Keep attributed the decline in R&D spending to artificial intelligence technology enhancing labor productivity, the financial report data clearly indicates that this change is highly correlated with the decrease in compensation costs due to employee turnover.

More importantly, this adjusted profit is not true profitability in the real sense. Under international financial reporting standards, Keep’s net loss in 2025 still reached 71.786 million yuan, although this was an 86.6% reduction from a loss of 535 million yuan in 2024, but the company has not escaped the state of actual losses, and the so-called profit resembles a “numbers game” on the financial statements.

User Loss of 8 Million in a Year

If the accounting profit achieved through cost reduction is Keep’s “superficial accomplishment,” then the significant loss of core users is its most fatal concern for future development. In China’s mobile internet industry, user scale is the underlying core asset of the platform, directly determining traffic value, monetization potential, and industry voice, yet against the backdrop of a growing national fitness market, Keep’s user scale has experienced a significant reverse shrinkage.

In 2025, Keep’s average monthly active users were 21.77 million, a year-on-year decline of 27.24%, with a loss of 8.15 million users within a year. At peak data, its monthly active users reached 36.39 million in 2022, with the number of lost users exceeding 14.62 million over just three years, nearly “halving.”

In stark contrast, the General Administration of Sports of China published the “2025 National Fitness Activity Status Survey Bulletin,” showing that the proportion of residents aged 7 and older who regularly participate in sports exercise reached 38.52%, up 4.6 and 1.3 percentage points from 2015 and 2020, respectively, with the number of urban and rural residents participating in sports exercise continuing to increase. While the industry overall is on the rise, Keep’s user scale is continuously declining, indicating serious issues with its product attractiveness and user retention capabilities.

The impact of user loss directly affects the core membership business, with Keep’s average monthly subscription members in 2025 at 2.74 million, a year-on-year decline of 13.29%, marking consecutive years of decline. In 2022, this figure was still 3.62 million.

Although the company has achieved an increase in single user value through operational optimization, with average monthly revenue per active user reaching 6.3 yuan, an 8.9% year-on-year increase, and membership penetration rising to 12.6%, the significant shrinkage of the user base means that the increase in single user value cannot compensate for the loss of scale effects, and the platform’s monetization efficiency and long-term commercial value are being continuously weakened.

Regarding the substantial user loss, Keep explained it as a strategic choice to proactively reduce customer acquisition costs and focus on core user operations, claiming that the user structure has achieved “overall fluctuation and structural optimization.” However, in an era where traffic dividends are fading, losing the support of a user base renders the so-called “structural optimization” akin to a source-less water or a tree without roots, ultimately leading to the continuous dismantling of the platform’s commercial barriers.

Stock Price Plummets by 90%

The situation of user loss and business cooling is also directly reflected in the performance of the capital market.

In July 2023, Keep went public on the Hong Kong Stock Exchange at an issuance price of 28.92 HKD, briefly spiking before entering a prolonged downward trend. In February 2025, founder Wang Ning released a letter to all employees on the tenth anniversary, proposing the “AllinAI” strategy and expanding the global footprint, which temporarily ignited enthusiasm in the capital market, with the stock price soaring by 32% in a single day, but this excitement was short-lived, with the stock price declining by 36% for the entire year of 2025. Entering 2026, the stock price decline continued, with a drop of 20% for the year, and as of March 26, the closing price was only 2.91 HKD, hitting a historic low, down 90% from the issuance price, with a current market value of only 1.5 billion HKD, far from market expectations at the time of listing.

The continuous pessimism in the capital market essentially reflects a deep skepticism regarding the effectiveness of Keep’s AI strategy implementation. In the first half of 2025, Keep launched its self-developed AI coach “Kaka,” focusing on personalized exercise guidance, and data from July 2025 indicated that its core AI daily active users exceeded 150,000, while the company’s overall daily active users stood at 4.23 million, with AI core users accounting for less than 4%.

Although Wang Ning once estimated during the 2025 first-half earnings call that the AI daily active users could exceed 1 million by the end of the year, this target was not fulfilled in the annual report, only disclosing that it had generated personalized training plans for a cumulative 1.3 million users, with the effectiveness of the AI strategy falling significantly short of expectations.

Undeniably, Keep has made initial progress in the AI field, with its self-developed sports health model completing underlying architecture upgrades, and the AI coach “Kaka” achieving full-scene coverage with voice guidance, diet management, and action assessment. AIGC has also reduced costs and increased efficiency in content production, but at this stage, the AI business is still in the investment and validation phase, lacking the capability for scalable revenue generation, making it difficult to become the core engine supporting the company’s development. Meanwhile, competition in the internet fitness sector is becoming increasingly fierce, with traditional fitness brands accelerating digital transformation and emerging technology companies continuously entering the market, making AI applications a standard in the industry. Although Keep’s AI strategy started early, it has yet to build an irreplicable technological barrier.

The profitability model achieved through “cost reduction” is destined to be unsustainable; once the benefits of contraction in marketing, R&D, and other areas are fully realized, if substantial growth in the core business cannot be achieved, Keep will once again fall into a profitability crisis.

For Keep, the first adjusted profit in 2025 is merely a sign of successful “blood-stopping,” not the end of development. In the context of the national fitness trend, the internet fitness sector still has immense growth potential, but the market will no longer pay for mere “numbers games.”

Only by abandoning short-term cost-cutting thinking, allowing the AI strategy to transition from “concept” to “implementation,” transforming technological advantages into actual drivers of user growth and revenue increases, and simultaneously reconstructing product competitiveness to regain lost user trust, can Keep truly establish its foothold in the industry competition.