KE - Earnings Call - Q3 2025

Transcript

Operator (participant)

Hello, ladies and gentlemen. Thank you for standing by for KE Holdings Third Quarter 2025 Earnings Conference Call. Please note that today's call, including the management's prepared remarks and question-and-answer session, will all be in English. Simultaneous interpretation in Chinese is available on a separate line for the duration of the call. To access the call in Chinese, you will need to dial into the Chinese line. At this time, all participants are in listen-only mode. Today's conference call is being recorded. I will now turn the call over to your host, Ms. Ting Li, IR Director of the company. Please go ahead, Ting.

Ting Li (IR Director)

Thank you, Operator. Good evening and good morning, everyone. Welcome to KE Holdings' Third Quarter 2025 Earnings Conference Call. The company's financial and operating results were published in the press release earlier today and are posted on the company's IR website, investors.ke.com. On today's call, we have Mr. Stanley Peng, our Co-founder, Chairman, and Chief Executive Officer, and Mr. Tao Xu, our Executive Director and Chief Financial Officer. Mr. Xu will provide an overview of our business updates and financial performance. Mr. Peng will share more on our strategic developments and innovative initiatives. Before we continue, I refer you to our safe harbor statement in our earnings press release, which applies to this call, as we will make forward-looking statements. Please also note that KE Holdings' earnings press release and this conference call include discussions on unaudited GAAP financial information, as well as unaudited non-GAAP financial measures.

Please refer to the company's press release, which contains a reconciliation of the unaudited non-GAAP measures to comparable GAAP measures. Lastly, unless otherwise stated, all figures mentioned during this conference call are in RMB. Certain statistical and other information relating to the industry in which the company is engaged to be mentioned in this call has been obtained from various publicly available official or unofficial sources. Neither the company nor any of its representatives has independently verified such data, which may involve a number of assumptions and limitations, and we are cautioned not to give undue weight to such information and estimates. For today's call, the management will use English as the main language. Please note that the Chinese translation is for convenience purposes only. In the case of any discrepancy, management statements in their original language will prevail.

With that, I will now turn the call over to our CFO, Mr. Tao Xu. Please go ahead.

Tao Xu (CFO)

Thank you, Mr. King. Thank you, everyone, for joining our Third Quarter 2025 Earnings Conference Call. In Q3, under the strategy of balancing skill and efficiency, we will further optimize our business structure, enhance operational and middle and back office efficiency through AI technology, and achieve the city-level profitability in both our home renovation and rental business before deducting high-quality expenses. Their combined contribution profit to the company's total gross profit reached a record high. The cost and expenses of our core business segments were further optimized. We also significantly enhanced the execution of shareholder returns with single-quarter share repurchase spending, reaching its highest level in the past two years. Regarding our overall financial performance in Q3, our total GTV was RMB 736.7 billion, remaining flat year-over-year. Total revenues reached RMB 23.1 billion, up 2.1% year-over-year.

Gross margin declined by 1.3 percentage points year-over-year to 21.4%. GAAP net income was RMB 747 million, down 36.1% year-over-year. Non-GAAP net income was RMB 1.29 billion, down 27.8% year-over-year. With that overview, I'd like to provide some details on operational and financial performance for each segment. Looking at our housing transaction services, we have been continuously enhancing the productivity and operational performance through the application of AI and other technologies, as well as in-depth operational optimization. For our in-home transaction services, we upgraded our AI to Hao Ke. As of the end of the third quarter this year, high-quality business opportunities identified through Hao Ke account for only a single-digit percentage of total potential leads, yet contribute over 50% of the transaction volume on our platform.

On the housing supply side, we launched innovations such as agent specialization modules, which agents are assigned to specialty managers, home listing, or server, also the buyer based on their expertise, as well as innovative services, including home staging and open house events. These efforts have enhanced the buyer conversion and the marketing and the sales through efficiency of the home listings. For our new home transaction services, we have also continuously iterated our AI agent, Tianji System, for intelligent operations and marketing, as well as AI assistant, Tianju. In terms of the financial performance, revenue from in-home transactions reached RMB 6 billion in Q3, down 3.6% year-over-year, and down 10.8% quarter-over-quarter. GTV was RMB 505.6 billion, up 5.8% year-over-year, and down 13.3% quarter-over-quarter.

The GTV growth outpaced revenue on a yearly basis, mainly due to a higher GTV contribution from in-home transactions facilitated by connect agents, for which revenue is recorded on a net basis. While revenue performance outpaced the GTV quarter-over-quarter, mainly due to the structural shift as revenue contribution from the rental brokerage services increased amid seasonal fluctuations, which have a relatively high take rate. The contribution margin of the in-home business was 39% in Q3, a decline of 2 percentage points year-over-year, primarily due to the relatively stable fixed labor costs amid the revenue decline. Essentially, the contribution margin declined by 1 percentage point due to the decline in revenue exceeding the fixed labor costs. Our new home GTV reached RMB 196.3 billion in Q3, down 13.7% year-over-year and 23.1% quarter-over-quarter.

Revenue from the new home transactions was RMB 6.6 billion in Q3, decreasing by 14.1% year-over-year and 23% quarter-over-quarter. Revenue performance was in line with GTV performance both year-over-year and quarter-over-quarter, reflecting our steady monetization capability in new home business. The contribution margin from the new home transaction services was 24.1%, down by 0.7 percentage points year-over-year due to an increase in variable costs resulting from our agent benefit improvement last year. On a quarterly basis, the new home contribution margin fell by 0.3 percentage points, largely due to higher variable costs and a smaller decline in fixed labor costs compared with revenue. For our home renovation and furniture services, we continued to strengthen our core capability to support long-term sustainable growth. On the product side, we successfully replicated our productized showroom model in multiple cities.

On the supply chain side, we expanded our centralized procurement categories and adopted localized sourcing standards and the selection process, further reducing the overall unit purchase price. To enhance delivery quality, we focused on improving construction quality, standardizing on-site management, laying the foundation for a unified system to excise construction site quality. In terms of the financial performance, revenue from our home renovation and furniture business was RMB 4.3 billion, remaining relatively flat year-over-year. Contribution margin for the segment reached 32%, up 0.8 percentage points year-over-year, primarily driven by the reduced procurement costs resulting from a larger proportion of centralized purchasing and the decreased labor costs resulting from enhanced order dispatching efficiency. Sequentially, the contribution margin remained relatively stable. For our home rental service business, on product front, our new Zero9 product has been launched in 10 cities, offering property owners diversified service options.

For unit sales and occupation, our improved operational efficiency through AI-powered housing condition assessment and intelligent pricing, while further promoting our quality-based traffic allocation rules to achieve faster housing turnover. In Q3, the conversion ratio of carefully run business opportunities to rental deals increased by more than 2 percentage points year-over-year. In terms of the operational management, we enhanced the productivity for the property managers and other personnel through the further refinement of the role specialization of labor, the integration of operational process, and the empowerment of AI technology. Regarding financial performance, revenue from our home rental services reached a record high of RMB 5.7 billion in Q3, up 45.3% year-over-year, driven by rapid growth in the number of rental units under management. By end of Q3, we had over 660,000 rental units under management, compared with over 370,000 in the same period of 2024.

The contribution margin for home rental services was 8.7%, up 4.3 percentage points year-over-year, and 0.3 percentage points quarter-over-quarter, largely driven by improved gross margin from our carefully run business. As we continue to refine the business model, we have adopted a net revenue recognition approach based on service fees for the certain newly signed properties, in line with the nature of the underlying service contracts. In Q3, our revenue from emerging and other services decreased by 18.7% year-over-year and 8.4% quarter-over-quarter to RMB 396 million. Now, moving to the four other financial metrics in Q3, including other costs and expenses, profitability, and cash flow. Our store costs reached RMB 663 million in Q3, decreasing by 5.8% year-over-year and 13% quarter-over-quarter, mainly due to the lower store rental costs.

Gross profit dropped by 3.9% year-over-year to RMB 4.9 billion. Gross margin was 21.4%, down 1.3 percentage points year-over-year. The decline was mainly due to the structural impact from a lower revenue proportion of in-home and new home business, which have relatively high contribution margins, as well as the decrease in contribution margin from the in-home business. This was partially offset by an increase in contribution margin from the home rental services. Gross margin declined by 0.5 percentage points quarter-over-quarter in Q3, mainly due to the structural impact as the revenue contribution of new home transaction service declined. In Q3, our GAAP operating expenses totaled RMB 4.3 billion, down 1.8% year-over-year and 6.7% quarter-over-quarter.

Notably, G&A expenses were RMB 1.9 billion, relatively flat year-over-year and down by 10.3% quarter-over-quarter, primarily attributable to the decreased bad appropriations and reduced share-based compensation expenses. Sales and marketing expenses were RMB 1.7 billion, down 10.7% year-over-year, mainly due to the lower personnel expense and reduced advertising and promotion expenses under the efficiency enhancement strategy. On a quarterly basis, the sales and marketing expenses were down 9%, mainly driven by a reduction in labor-related costs. Our R&D expenses were RMB 648 million, up 13.2% year-over-year and 2.3% sequentially, largely driven by higher personnel expenses. In terms of the profitability, GAAP income from operations totaled RMB 608 million in Q3, down 16.4% year-over-year and 42.6% quarter-over-quarter. GAAP operating margin was 2.6%, dropping by 0.6 percentage points from Q3 2024 and 1.4 percentage points quarter-over-quarter.

The non-GAAP income from operations totaled RMB 1.17 billion, decreasing 14% year-over-year and 27% quarter-over-quarter. Non-GAAP operating margin was 5.1%, down 1 percentage point from Q3 2024, mainly due to the decline in gross margin. Non-GAAP operating margin was down 1.1 percentage points from the previous quarter, mainly due to the increase in operating expenses ratio sequentially. GAAP net income totaled RMB 747 million in Q3, down 36.1% year-over-year and 42.8% quarter-over-quarter. Non-GAAP net income was RMB 1.29 billion, falling 27.8% year-over-year and 29.4% quarter-over-quarter. Moving to our cash flow and balance sheet, we generated net operating cash inflow of RMB 851 million in Q3. New home DSO remained at a healthy level with [Foreign language] 54 days in Q3.

In addition to spending approximately $281 million in share repurchase during Q3, our total cash liquidity, excluding customer deposits payable, remained at around RMB 70 billion. Facing the short-term business challenges brought by external fluctuation and internal strategic transformation, we support and reward our shareholders through consistently active share repurchase to improve the efficiency of the capital operations. From the fourth to third quarter of this year, we spent $139 million, $254 million, and $281 million on share repurchase, respectively, with the cumulative amount of approximately $675 million in this year, up 15.7% year-over-year. As of the end of Q3, the number of repurchased shares accounts for about 3% of the company's total issued shares at the end of 2024.

Since the launch of our share repurchase program in September 2022, we had repurchased around $2.3 billion worth of shares as of the end of September this year, accounting for about 11.5% of our total issued shares before the program began. We have made progress in Q3 this year in proactively optimizing our business structure, strengthening technology empowerment, and enhancing shareholder return. Our forward-looking layout of the home renovation and furniture services and home rental services have both achieved profitability at the city level before deducting high-quarter expenses in third quarter. The AI capabilities have shown initial results in driving the business development and improving the work efficiency of the service provider and the middle and back office personnel. We are also fulfilling our shareholder return commitment with greater intensity, repurchasing $281 million in a single quarter, increasing 38.3% year-over-year.

As the industry enters a new stage of high-quality development, we are taking initiative in building a residential services ecosystem. With our combination of technological innovation, anti-cyclical business portfolio, and a highly efficient and well-structured operating system, we are well-positioned to deliver great value to both customers and investors. Thank you. Next, I would like to turn the call to our Chairman and CEO, Stanley.

Stanley Peng (Co-Founder, Chairman, and CEO)

Thank you, Paul, for sharing our business and financial developments for the third quarter. We are strategically shifting our growth engine from scale to efficiency. Today, I'd like to highlight some innovative initiatives we have rolled out across businesses to advance this shift. First, in terms of our core business transaction services, externally, we see new demand from both buyers and sellers under the new norm for China housing market. Home sellers expect stronger marketing capabilities from us.

Buyers are counting on us for customer-oriented insights to support their decision-making in areas such as timing, asset planning, and listing comparisons. These trends place due requirements on our traditional agent skill model, and agents who are great at supporting both buyers and sellers are extremely rare. Since mid-year, we have been working to restructure our capabilities across both buyer and seller agents. In Shanghai, we piloted a seller and buyer agent specialization mechanism to enhance our marketing and operating excellence on the home seller's agent side first. The mechanism redefines organizational roles, commission structures, and performance initiatives, and offers supporting tech products. This, in turn, allowed buyer-side agents to prioritize quality listings and improve transaction conversion. The underlying logic is that high-quality home listings by engineers not really made. They require skilled agents to master market analytics, pricing, property staging, owner engagement, and decision-making.

Precision marketing, second, eventual equality drives customer acquisition. Superior listings inherently attract more serious buyers, driving transaction speed and our brand reputation, which in turn attracts better talent to join us. Therefore, we did several things to implement this. First, we adjusted our organizational structure and incentive mechanisms. We shifted some senior agents into [hybrid] roles that combine management and home seller-focused responsibilities, giving them the authorities to form and lead their own teams dedicated to listing management. Under the ACN commission allocation mechanism, we raised the selling agent share from 40% to over 50%. We are maximizing incentives for top-performing agents to focus on marketing high-quality home listings. This group of home seller-focused agents can earn around 25% more than before, assuming our market share remains stable.

To mitigate potential pressure on buyer's agents, we reduced the mandatory 10-row commission split, raised the minimum commission for selling agents, and offered extra incentives for selling high-score listings. Second, we provided agents with systematic support and digitalized products to help them manage listings. In the past, homeowners' relationship management, listing presentation, and marketing relied on agents' personnel experience. That made it hard to replicate and scale. We have built an AI-powered listing score system that captures and qualifies the know-how required in six key areas: home listing maintenance completeness, homeowner engagement depth, property condition, for example, renovation recency, listing cross-channel marketing performance, AI-powered pricing competitiveness, buyer's interest, for example, the listing's online/offline viewings. These metrics help agents clearly understand what defines a high-quality listing and how to better present and market homes. Home buyer agents can also focus on selling nice-score listings to drive better sales conversions.

In terms of results, in September, high-scoring listings accounted for more than 75% of transactions. Our average market coverage in Shanghai hit a record high in Q3, increased 1.2 percentage points year-over-year and 2.6 percentage points quarter-over-quarter. The experience of homeowners looking to sell quickly also improved. Many homeowners reached out to us proactively to learn how to raise their listing scores. Buyers also naturally prefer high-scoring listings, creating a positive cycle that benefits everyone involved. The home seller buyer-side agent specialization in Shanghai is an important initiative designed to meet the changing needs of our customers and marks a milestone in our shift from scale to efficiency. We will continue to track its progress and explore new initiatives on the home buyer's agent side. In addition, we tried innovative approaches to make our new business more efficient.

For example, in our home rental business, Q2 marked the first time we excluded headquarter costs from break-even at the city level, and Q3 is expected to contribute over RMB 100 million in profits. Carefree Rent, our decentralized long-term rental business, housing businesses inherently face challenges, including relatively low average selling prices, non-standardized products and services, extensive service coverage, and high maintenance costs, traditionally requiring heavy manpower and variable cost investment for scaling and operating. This sector has struggled with economics of scale industry-wide, with no established base practices yet. As newcomers, we embraced this as an opportunity to build an AI-native operation from inception, enabling parallel developmental business capabilities, frontline operations, and AI intelligence. Through our organizational restructuring, process optimization, and AI driving and products, we are pioneering an air traffic efficiency, economically sustainable model. Early results demonstrate significant improvements, offering valuable insights for our other platform businesses.

I'll walk you through three major AI-driven black stores across different dimensions. First, AI has been fully integrated into our rental services business, enabling end-to-end intelligent decision-making and business operations. For rental unit signups, AI now powers critical processes, including property lead identification, personnel management and deployment, property evaluation, pricing strategies, and homeowner communication. For example, previously, personnel management and operational relies heavily on experience level with supervisors deciding which agent would be responsible for which area. Now, through AI-driven grid management supported by our unique dynamic domain data and modeling capabilities, AI can make data-driven determinations. It evaluates factors such as the number and quality of property leads, local supply-demand relationships, and personnel capabilities models. Based on this dataset, it determines the optimal personnel assignments, regional coverage, and organizational structure. AI can simulate up to 90,000 design scenarios per minute, automatically generating the most efficient staffing and operational strategies.

This has greatly improved how we allocate our service personnel's deployment, configuration, and operational scope. We also use AI to guide and execute our core business strategies and daily helping us move forward fully in intelligent operations. For rental unit signup, we roll out an AI-powered rental unit signup system that uses real-time data and algorithms to predict market demand, property inventory, and price trends. It generates automated signup strategies and dynamic pricing recommendations, delivering tailored plans for each property through adaptive decision models. As market conditions change, such as customer demands, property inventory, and pricing, AI can guide our operations team to make timely adjustments. For example, when there is an oversupply of three-bedroom units in a certain area, the system automatically triggers price controls and signup restrictions. When unit types are in short supply, AI reactivates dormant property leads.

Our upcoming AI cloud buffer will also automatically contact homeowners of these reactivated properties. In Linyi where we began pilot operations in August, our workforce decreased by 10%, while new rental signup units grew up 10% even in the off-peak season. For rental unit leasing, our AI inventory management system frequently monitors inventory and checks over managing high-risk or low-maintenance properties. It dynamically adjusts pricing and targeted discounts while optimizing traffic to speed up leasing. In Q3, these capabilities accelerated the lease out of 350,000 units across 11 cities with 90% price adjustment adoption. This effort generated over RMB 100 million in nationwide cost savings. Second, we use AI and technology to solve the industry's long-standing problems with non-standardization, enabling high-quality, scalable growth. The home rental industry has several characteristics. Home listings are scattered, and each home has different and complex internal conditions, making the products non-standard.

Service providers are many, and their levels vary, so the workforce is also non-standard. Market price fluctuates, and traditional pricing relies on frontline staff's on-site judgment, leading to non-standard pricing. Operational processes are mostly offline and complex, making sales strategies and service execution non-standard as well. There are the traditional constraints of the industry, but with the progress of AI, we see challenges to achieve both standardization and personalization at the same time. At the property quality and risk assessment stage, we have achieved human-AI integration, with AI now leading the entire unit's signup workflows. Our AI property evaluation assistant uses visual recognition and multimodal analysis to intelligently capture indoor features, assess property conditions, and evaluate potential risks. It also incorporates market data to generate intelligent AI-driven pricing recommendations.

Beyond analyzing photos, the system can interpret property attributes holistically, helping address challenges such as consistent product standards, varying personnel capabilities, and pricing accuracy. In the homeowner communication phase, we launched the AI negotiation assistant. This tool packages AI-driven property assessment, dynamic pricing, and competitive market data into tailored home signup strategies and negotiation scripts, helping our service providers communicate and negotiate with homeowners more effectively. This provides a more professional and friendly experience for our clients, equipping new service providers with the tools they need to grow quickly and learn how to address non-standard sales issues. We piloted its feature in Linyi, and unit signup productivity rose by over 10+ percentage points in Q3 compared with Q2, ranking number one nationwide. Third, we achieved a leap in efficiency by adopting different AI applications.

During the signup stage, our AI review system has replaced manual reviews, enabling fully automated risk control. As of September, the AI review function covered 11 cities, processing each case in just 20 seconds on average, making a 64% efficiency gain, saving more than 33,000 work hours and intercepting more than 16,000 risky properties. In the leasing stage, we use AI to power content lead marketing, expanding lead generation while reducing labor needs. AI intelligently analyzes and identifies high-quality leads, enhancing leasing efficiency. The AI-driven operational system in our home rental services has enabled us to see the possibility of scalable yet personalized services for previous fragmented, non-standardized demands, demonstrating the potential for traditional industries to overcome these economics of scale through technological innovation. We now integrate AI across our entire home rental services process, and I'm replicating this system across 13 key cities.

Only through continuous innovation can we navigate the industry cycle. By implementing home buyer-seller agent specialization and AI-driven home rental operations, we have forged a new path that re-engineers workflows through technology and fuels scale through efficiency. Moving forward, we will deepen AI integration across business scenarios to advance both service providers' capabilities and consumer experiences. As China's housing service industry undergoes this next evolution, we are afforded a historical opportunity to further its transformation, guided by our commitment to technology power, high-quality growth, and its potential to unlock infinity possibilities for modern living services. This concludes my prepared remarks for today. Operator, we are now ready to take questions.

Operator (participant)

Thank you. As a reminder, we only accept questions on the English language line. For the benefit of all participants on today's call, please limit yourself to one question, and if you have additional questions, you can re-enter the queue. If you're going to ask a question in Chinese, please follow with an English translation. Your first question comes from John Lamb with UBS.

John Lamb (Analyst)

呃，喂，呃，请问可以听到吗 [Foreign Language]？

Tao Xu (CFO)

可以。[Foreign Language] [crosstalk]

John Lamb (Analyst)

home business going forward [Foreign language]? Um [Foreign Language], so let me translate my questions. So for the new home business, in the past, the company has been achieving, or outperforming, the market, in terms of the Alpha, but then it seems that the magnitude of the Alpha has been diminishing. May I know what's the reason why, and also how should the investor look at the company new home business growth potential? Thank you.

Tao Xu (CFO)

Thank you, John. Although the near-term performance of our new home construction business has been affected by the market volatility, we remain confident in its ability to outperform the market in the long run. China's new home market has gradually matured in the past two years, with supply-side risks steadily easing. Against this backdrop, we have shifted from a cautious approach to a more growth-driven strategy. Our new home construction business has significantly outperformed the broader market in past few quarters until this second quarter, with a higher brokerage penetration in the industry. Our broader housing construction service network and more collaborative projects. In this Q3, our year-over-year growth narrowed relatively to the market, mainly due to several factors. First, customers on our platform often look at both new and existing homes before making a purchase decision.

Recently, the prices of existing homes have been considerably more attractive than the prices of comparable new homes, leading both first-time buyers and home upgraders to choose existing homes. Second, this is a basic fact. The platform's new home transaction had a relatively higher base in last Q3, as many policy-driven new home subscriptions in Q2 were transacted in Q3, causing a timing mismatching with the market data. Third, of course, it is important to note that in recent years, our new home business has grown rapidly from a lower base as we made significant gains in brokerage penetration. The scale of our collaborative projects and our sales through network and capabilities, we estimate the brokerage channel penetration ratio in the new home market has grown to over 50% this year, from approximately 30% a few years ago.

In cities we operate in, the coverage of our collaborative project has expanded to over 70%, from roughly 39% in 2023. To achieve further growth in a higher base, we have several key opportunities. First, we plan to expand into more cities and broaden our target market. Second, brokerage channel penetration in China still lags behind developed markets, leaving ample room for growth. Third, we leverage refined operation management to enhance the service capability for the new home customers and the sales through efficiency, as well as improve our coverage and sales through capability for high-end products. Now, let's take a closer look at the details. First, we are piloting lighter product offerings to tap in some lower-tier cities through what we called B+ products. Our platform business still has over 150 prefecture and county-level markets now yet to be covered.

Building on our commitment to authentic listings, the B+ pilot equips local brokerage stores and agents in more cities with the system capability, traffic support, and commercialization tools. This lighter operation approach enables more flexible collaboration on home listings and sales, and the new home sales with our channel partners. As of September 2025, our B+ business has been piloted in four cities, and we plan to expand to over 30 cities by the end of the year, unlocking additional market opportunities. Second, we see the room to grow our sales opportunity with collaborative projects. Under the customer end, we will optimize content development and operational strategy for our new home business to reach more buyers and increase the conversion rates. Under the customer end, we will iterate our partnership models and product offerings to developers.

Third, both supply and demand in new home markets are increasingly shifting towards the home upgrade projects. On the supply side, we will more precisely identify these projects and boost their exposure to both agents and customers, with matching suitable agents to these upgrade projects and direct more customer traffic to them, creating a closed loop among homes, agents, and customers. This approach will also help agents strengthen their sales capability for upgrade products and narrow the price gap between the platform average new home unit and the broader market. Thank you.

Operator (participant)

Your next question comes from Griffin Chen with City.

Griffin Chen (Analyst)

yeah, I'm going to translate my question [Foreign language]. So this is Griffin from City Property Team. How did the leasing service business manage to turn last year's losses into the operating profit by third quarter this year? And what opportunity remains for the improvement going forward? Thank you.

Tao Xu (CFO)

Yeah, thank you, Griffin. The possibility of our home rental services improved significantly this year, excluding [high-cost] allocations. City-level operating profits became in Q2 and became profitable into Q3. First, we benefited from economies of scale from rapid growth in both scale and revenue. The total number of matched units exceeding 660,000 by the end of Q3, up 75% year-over-year. Revenue from our home rental service business returned to be RMB 5.7 billion in Q3, up 45.3% year-over-year. The contribution profit from our home rental services also rose significantly to nearly RMB 500 million in Q3, up 186% year-over-year, with a contribution margin of 8.7%, up 4.3 percentage points year-over-year. On one hand, the light access model of our carefully run business has given us a higher margin, lower risk rental structure.

Starting in this Q3, the revenue from newly added rental units and renewed listing units under Carefree Rent has been accounted on a net basis. In Q3, rental units under the net revenue accounting method made up 25% of the total units under management, up 10 percentage points quarter-over-quarter, contributing approximately RMB 470 million in revenue. This structural shift drove an RMB 130 million increase in Carefree Rent Q3 contribution profit and lifted its contribution margin by 3 percentage points. At the same time, 2025 has been a year of improving operation efficiency. The streamlined and highly efficient operation have driven the reduction in several cost ratios, adding about RMB 170 million to contribution profit and increasing contribution margin by roughly 1.5 percentage points. Excluding rental costs recognized on a gross basis, the main costs of Carefree Rent are labor costs, channel costs, post-rental installation, and default costs.

The improvement was mainly driven by the optimized operation labor costs. In Q3, the average monthly number of units managed per property manager exceeded 130, compared with over 90 in the same period last year. In the fourth quarter of this year, average monthly efficiency in unit size and occupancy rose by approximately 10% and 28% year-over-year, respectively. The default cost ratio declined by 0.1 percentage points, benefiting from our strong leasing capability. In Q3, initial leasing success rate improved by 0.9 percentage points year-over-year. So far this year, contribution profits from our home rental business segment have grown much faster than operating expenses. These expenses mainly comprise high-cost and city-level staff compensation and R&D, with a quite low expense ratio. A series of operating management tools have consistently improved the productivity of our middle and back office personnel.

The average number of units under management by each middle and back office personnel rose by 7.5% year-over-year, while the overall operating expense ratio declined year-over-year. In the coming years, there is a significant room to continuously improve the contribution margin in our carefully run business. The key drivers will be the continuous growth potential of the rental unit scale of the Carefree Rent and the ongoing improvement of our operational efficiency. From a per-unit optimization perspective, we are diversifying our channels for renting out our property to reach broader tenant demographics, increasing the share of our in-house rental occupancy team and reducing reliance on the concentrated brokerage channels. This is expected to lower the per-unit channel cost ratio. In addition, labor costs remain a large part of per-unit UE, and there is still room for further reduction of the cost ratio.

We see the potential to nearly double the number of units managed per property manager, moving towards an average of over 200 units per person. Furthermore, we will keep exploring and expanding diverse value-added services with the home rental ecosystem. We will continue to invest in AI and online digital capability within our home rental service, while other operating expenses should stay relatively stable. As the business continues to scale and we further optimize per-unit UE, we expect our home rental service to maintain a strong operating leverage in the year ahead. Thank you.

Operator (participant)

Your next question comes from Jiang Xiao with Barclays.

Jiang Xiao (Analyst)

Thank you very much for taking my questions. Good evening. My question is around your renovation business. You have done very well in cities like Beijing and Shanghai. I was just wondering, for you to do well in those cities, is that because you have high market share with your Lianjia brand in those cities? Do you think that is a key reason? Do you think for cities outside Shanghai and Beijing, how would you kind of motivate your agents to cross-sell or to sell the renovation business when you do not have such a high market share? Thank you so much.

Tao Xu (CFO)

Thank you, Jian Xiao. First of all, it is important to note that the home renovation market in second and third-tier cities represents a critical long-term growth driver for our future home renovation business, carrying irreplaceable strategic value. From a market fundamental perspective, compared to the first-tier cities, the cost of purchasing a similar size of property is much lower in small cities. Based on the latest data from our platform, the average price of a single home in Beijing and Shanghai is around RMB 4 million, versus just over RMB 1 million in other cities. This price gap presents a meaningful opportunity as customers in second and third-tier cities can allocate a relatively larger budget for the home renovation. In 2024, we recorded approximately 1 million existing home transactions outside Beijing and Shanghai.

In these cities, home renovation contract orders generated through our agent network only accounted for around 30% of overall home renovation contract orders. Our conversion rate from existing home transaction to home renovation contract in these cities was just less than 5%, compared to over 20% and 10% in Beijing and Shanghai, respectively. Our strategic rationale is clear. Larger scale expansion into additional cities will only begin once the home renovation business underlying operational capability are matured. The module has been fully proven in core cities. Therefore, our resources are highly concentrated in core cities at this moment. We have not yet made a big effort to drive traffic for our home renovation business through our Lianjia agent channel in the second and third tier cities so far. This approach is to ensure that every step of our goals is solid and sustainable.

Meanwhile, we put in place a multidimensional systematic operational framework to engage with and motivate our Lianjia agents. It includes three components. First, we aim to deepen our operation teams' understanding and expertise in home renovations. Our operation teams have also shared the knowledge and proven operational capability to connect store owners and agents, fostering an ecosystem marked by professional collaboration and shared competency. Second, we rolled out innovative incentive programs to build online brand promotion metrics. By offering incentives such as bigger coins, we encourage more connected store agents to visit our offline home renovation stores and showcase our service through the short video, which they will also upload to the leading social media platforms such as Douyin. Since the launch of this program in the late April of this year, more than 30,000 agents in over 30 cities have uploaded over 50,000 short videos.

This has cultivated a positive environment of full participation and widespread promotion. On top of improving agent capability, we are leveraging AI to boost the contract conversion efficiency. Using AI, we access key attributes of the property within the store owner's coverage area, such as property age, layout, condition, and assess quantitative scores. This allows us to accurately identify high-scoring homes with a higher likelihood of generating home renovation business. Feedback from the pillar cities has been extremely positive. While high-scoring homes constitute only low single digits of the total home renovation lease, they contribute to over 20% of preliminary home renovation contracts, underscoring AI's value in boosting our operation efficiency. In Q3 this year, our home renovation leads from our Lianjia agent channels achieved year-over-year growth, and the lead-to-contract conversion rate increased compared with last year's average.

In the short term, our approach for the home renovation business remains relatively conservative. In the long run, once our home renovation service meets our established high standards across customer experience, product compatibility, and delivery quality, we will initiate a more proactive traffic diversion strategy through Lianjia agent channels in the cities outside Beijing and Shanghai. Thank you.

Operator (participant)

Your next question comes from Timothy Zhao with Goldman Sachs.

Timothy Zhao (Analyst)

Great. Good evening, management team. Thank you for taking my question. My question is about your cost and expenses. Could you further elaborate what are the measures for the company to control costs and any effects or outcomes that you have seen so far, and what we should expect from this cost and expenses line going forward? Thank you.

Tao Xu (CFO)

Yes, thank you, Timothy. Under the strategic guidance of operational efficiency enhancement, all businesses have already implemented a series of optimization measures and achieved [feasible] results. Now, I'd like to elaborate on the cost reduction, achievement of each business line, and overall operating expenses in the third quarter of 2025. For our existing home transaction services, we continue to boost the productivity of our Lianjia team, and organizational optimization has driven a notable decline in labor cost. Organizational optimization has directly led to a cost reduction, with the fixed labor cost in Q3 decreasing by more than 20% compared with the peak in Q4 last year. The labor efficiency has been continuously improved. For new home transaction services, we have both streamlined fixed labor costs and the variable cost structure through streamlining the organizational structure of new home operation team.

We have achieved a reduction of more than 40% relatively in relevant fixed labor cost compared to the peak in Q4 last year. On the variable cost side, the gross profit margin per project has been steadily increased by focusing sales strategy to maximize unit sales per single housing project. The commission speed of Lianjia channels has decreased by more than 1 percentage point from the peak in Q1 this year. For our home renovation and furniture business, we have effectively lowered the material cost through supply chain integration. By streamlining partner brand selection and SKU counts, we have achieved significant cost savings in procurement. Our centralized purchasing category has expanded from 4 as of Q2 to 13 as of Q3, covering core categories such as wooden doors, drawing, and tiles. The procurement unit price of some products has decreased by over 20%.

The effectiveness of the cost optimization has been reflected in the financial reports, with the proportion of material-related costs as a percentage of revenue in Q3 decreasing by about 1 percentage point compared to last year's average. For our home rental services, cost reduction has been driven by both technological improvement and business model refinement. We have improved the efficiency of the rental housing channel management through AI employment and task specialization of the service providers. The proportion of operating labor cost to revenue in Q3 decreased by around 1 percentage point year-over-year. For store cost, we have reduced fixed expenses through the refined management and closed underperforming stores. The number of active Lianjia stores has been decreased from around 500-600 as of Q4 last year to less than 500-200 at the end of Q3, a decrease of around 8%.

Meanwhile, we have actively promoted the rent negotiation with existing Lianjia store owners and achieved average rent reduction of over 10%. Regarding the control of the operating expenses and R&D investments, for G&A expenses, we have achieved efficient cost control through the organizational optimization. On a non-GAAP basis, the G&A expenses of the home renovation business have decreased by more than RMB 100 million compared to the peak in Q3 last year. This was mainly due to the adjustment of the organizational structure. The headquarters' G&A has also been optimized based on the market conditions. For sales and marketing expenses, both marketing spending optimization and the improvement of the labor efficiency have been implemented.

On a non-GAAP basis, the sales and marketing expenses of the housing transaction business have decreased by around RMB 90 million compared to the peak in Q3 last year, mainly through the optimization of the advertising and the marketing placement. The related advertising and promotion expenses have declined by more than 20% compared to the peak in Q3 last year. The sales and marketing expenses for home renovation business have decreased most significantly by more than RMB 100 million compared to the peak in Q3 last year. The core driving factors include AI technology enhancing the operation efficiency of the designers and other front-end staff, as well as organizational optimization that improved the workforce structure. For R&D expenses, the non-GAAP basis, the expenses in Q3 increased by around RMB 79 million year-over-year, as the scale of R&D team has expanded steadily.

As of Q3, there were more than 2,300 R&D-related personnel, an increase of more than compared with Q3 last year, among which the number of AI-related R&D personnel exceeded 600, doubling compared to the same period last year. R&D resources continue to be tilted towards the core areas, with R&D investment related to AI in Q3 exceeding RMB 150 million, nearly doubling compared to the same period last year. Our operational efficiency enhancement strategy has a clear execution path. We firmly believe that with the market environment stabilized, our continuous operation optimization will fully release the operating leverage effort. Thank you.

Operator (participant)

We are now approaching the end of the conference call. I will now turn the call over to your speaker today, Mr. Ting Li, for closing remarks.

Ting Li (IR Director)

Thank you once again for joining us today. If you have any further questions, please feel free to contact [KE Holdings'] investor relations team through the contact information provided on our website. This concludes today's call, and we look forward to speaking with you again next quarter. Thank you and goodbye.