More than 16,000 high-risk housing listings were intercepted, saving over 100 million yuan in costs. Beike steps up its "AI game".

On November 10th, Beike (NYSE: BEKE; HKEX: 2423) released its third - quarter earnings report. The data shows that in the third quarter, Beike's total transaction volume (GTV) reached 736.74 billion yuan, and its net revenue reached 23.05 billion yuan, a year - on - year increase of 2.1%. The gross profit margin was 21.4%, a year - on - year decrease of 1.3 percentage points. The GAAP net profit was 747 million yuan, a year - on - year decrease of 36.1%. The Non - GAAP net profit was 1.286 billion yuan, a year - on - year decrease of 27.8%.

Although affected by the overall downward trend of the real estate industry, Beike's profit level has declined, but potential growth space can still be seen in various businesses. In particular, the high - efficiency replication of the "One Body, Three Wings" model from key first - and second - tier cities to other cities will bring future market growth. In addition, the empowerment of AI technology in business and the middle - and back - office has brought higher labor efficiency and mature, separately - exportable light - asset products, providing Beike with another path for platform expansion besides "opening stores".

Beike

01  

How to challenge the impossible and achieve continuous growth?

Currently, the real estate industry has generally shifted from "pursuing growth" to "focusing on quality and efficiency". However, Beike still has new growth potential.

The basic logic of Beike's performance growth can be divided into two levels, corresponding to two different time periods:

The first level is the structural growth brought about by business layout. This stage started in July 2023 when the "One Body, Three Wings" strategy was proposed. The capabilities of the second - hand housing and new - home transaction businesses, as the basic business segments, have empowered other businesses. Through the reuse of human and store resources, the home improvement and home furnishing and housing rental businesses have achieved rapid growth, thus realizing the overall growth of the company's performance.

In the third quarter of 2023, Beike's overall revenue was 17.81 billion yuan. Among them, the revenue from the second - hand housing business was 6.3 billion yuan, the revenue from the new - home business was 5.9 billion yuan, the revenue from the home improvement business was 3.2 billion yuan, and the revenue from the rental business was 2.4 billion yuan. Calculated, the revenue from non - real estate transaction service businesses accounted for about 31.4%.

In the third quarter of 2025, in the 23.05 billion yuan of revenue, the revenue from non - real estate transaction service businesses accounted for 45%, becoming an important support point for Beike's performance curve.

Among them, the rental service business turned from a loss last year to a profit in the third quarter of this year, driving the proportion of the profits contributed by the home improvement and rental businesses to the gross profit to reach a record high of 38%.

After more than two years of layout, relying on the platform's capabilities, Beike has extended from the transaction link to the front - and back - end service processes. Through the in - depth integration of the service value chain and the networking of transaction services, it has achieved its growth goals.

If the core of the first stage lies in the improvement of the efficiency of human and stores and the synergy among various businesses, then in the next second stage, the core logic of Beike's growth will shift from the improvement of the platform's internal capabilities and efficiency to the structural opportunities in the external market.

At the third - quarter earnings conference, Beike's management mentioned: (Regarding the future new - home business) We have several opportunities. First, we will further expand into more cities to increase the scale of the target market. Second, there is room for further improvement in the market's channel penetration rate compared with developed countries. Third, through refined operation and management, we will improve our service capabilities and sales efficiency for new - home customers, as well as our ability to cover and sell high - end products.

Beike plans to penetrate into lower - tier cities through a lighter "B+" product, using traffic and commercial product tools to provide systematic capabilities to real estate brokerage stores and brokers in lower - tier cities.

As of the end of the third quarter, Beike's "B+" business has been piloted in 4 cities. By the end of 2025, it plans to enter more than 30 cities, driving business performance growth through market space expansion.

At the same time, as key cities shift to the improvement - oriented market, Beike's new - home business also plans to allocate more resources to the matching of brokers, customers, and housing sources for improvement - oriented projects. On the one hand, it will improve the pertinence of capabilities and transaction efficiency. On the other hand, it will increase the coverage ratio of cooperative housing projects in the cities.

In the home improvement and home furnishing business, next, Beike will focus on aligning the capabilities in second - and third - tier cities with those in first - tier cities. For example, in Beijing and Shanghai, the number of home improvement referral contracts from Beike's second - hand brokerage business is twice the average of second - and third - tier cities. As long as the mature models in first - tier cities are replicated in other cities, new growth can be achieved.

Another factor driving business growth is Beike's AI capabilities.

Beike is a company that deeply applies AI. In the transaction business chain, on the housing source side, Beike uses the "AI Housing Source Maintenance Assistant" to help brokers solve the problem of insufficient housing source maintenance due to limited energy. On the customer source side, the broker assistant "Laike" helps brokers activate customers from customer acquisition, house selection to answering questions, improving the brokers' professional service capabilities and the matching efficiency of housing sources and customer sources. In the signing process, Beike also has a digital partner "Xiaoyi" to complete signing quality inspections and automatic follow - up after signing.

In the rental business, through the collaborative operation of "AI + human", using AI to manage and operate housing sources, the labor efficiency of asset management managers has increased significantly. The data publicly released by Beike shows that in the third quarter of this year, the average number of housing sources managed per asset management manager per month exceeded 130, compared with more than 90 in the same period last year. In the first three quarters of this year, the average monthly labor efficiency of housing acquisition and rental in Beike's rental business increased by about 10% and 28% year - on - year respectively.

In the future, Beike plans to use AI capabilities to increase this data by more than 50%, reaching more than 200 housing sources managed per person on average, thereby driving the performance to reach a new level rapidly.

To sum up, Beike's future actions have two directions: expanding downwards to lower - tier markets externally and relying on AI internally.

02  

What can AI do?

Some enterprises only regard AI as a data analyst, but in Beike, AI is an all - process operation officer.

For example, AI has played a very important role in the rapid growth and profit turn - around of Beike's housing rental service. As of the end of September, the AI review function has covered 11 cities, with an average processing time of 20 seconds per case, a 60 - fold increase in efficiency, saving more than 33,000 hours of working time and intercepting more than 16,000 high - risk housing sources.

In the second - hand housing transaction business, in the first three quarters of this year, the high - quality business opportunities identified by Beike's AI product "Haoke" accounted for only a single - digit percentage of the platform's total potential business opportunities, but contributed more than 50% of the platform's transaction volume.

In the home improvement and home furnishing business, Beike uses AI to identify "high - score labeled housing sources" with a high probability of transaction. The pilot results show that the high - score labeled housing sources only account for a low single - digit percentage of the total housing sources entrusted for home improvement leads, but contribute more than 20% of the preliminary home improvement contract signings.

That is to say, Beike has already used AI to make decisions in the service process, and it has also proved that AI is more efficient and effective in handling problems in some links.

Especially in the rental business, which involves a large number of service details and non - standard scenarios, in fact, AI has no longer been just a "work assistant" but has been upgraded to "AI - driven operation" and has become a central part of the platform system.

What can AI do? Beike's answer is very clear. Taking the rental service business as an example:

In the housing acquisition stage, AI can not only perform "housing management" actions such as housing source lead mining and housing source evaluation, but also "manage people". Regarding housing sources, AI will independently guide the business to adjust the housing acquisition structure. For example, if there are too many three - bedroom houses in a certain area during a certain period, AI will automatically control the prices and ban the acquisition of housing sources exceeding the limit. At the same time, for housing sources with insufficient supply, it will automatically retrieve and analyze from silent housing source leads. Regarding personnel, Beike's AI will consider factors such as the number of housing source leads, housing source quality, supply - demand relationship, and personnel ability models in a certain area to plan personnel, regional, and organizational settings. It can simulate 90,000 design plans in one minute and automatically output the optimal personnel distribution and operation strategies.

In the housing rental stage, AI will conduct high - frequency inventory of housing source stocks, identify and directly manage high - risk and low - maintenance housing sources, adjust housing source prices dynamically, offer precise discounts, and allocate traffic to accelerate the rental of housing sources. A total of 350,000 managed housing sources have been rented out, accounting for 73.5% of all externally promoted housing sources. The adoption rate of AI - adjusted prices can reach 90%. At the housing rental end, AI - enabled intelligent management and precise marketing for rental have saved more than 100 million yuan in annual costs.

Within the Beike system, AI is demonstrating stronger operational systematization capabilities and a better understanding of housing sources. With the "second - level" housing source inventory speed and high - frequency, real - time analysis, it is changing the decision - making mode that used to rely on human anticipation and historical experience.

After the in - depth use of AI, Beike is challenging new management radii and scale limits.

Peng Yongdong, co - founder, chairman, and CEO of Beike, said: "The new - type AI - driven operation system in the rental business has shown us the possibility of serving the seemingly scattered and non - standard demands in a large - scale and personalized way, enabling the traditional industry to break through the 'diseconomies of scale'."

But Beike's AI is not intended to replace humans. Its essence is to deeply reconstruct the transaction and service chains, allowing humans to do personalized, warm - hearted, experience - based, and trust - building things, while AI conducts monitoring, analysis, prediction, and scheduling to extend human service capabilities, forming an "AI + human" collaborative operation system.

For Beike to achieve this, there is a very important prerequisite: a deep understanding of the industrial chain. It has made sufficient preparations in advance to disassemble the service chain, conduct extremely professional role - division, accumulate clean and structured data as fuel, enable human capabilities to be integrated into the platform, let the human role return to 'human nature itself', let the data system be rooted in scenarios, rather than asking humans for data and then making AI act like humans.