Waterdrop - Earnings Call - Q4 2024

Transcript

Tracy Li (Head of Investor Relations)

Good morning, everyone. This is Tracy Li from Waterdrop Investor Relations. It's my pleasure to welcome everyone to Waterdrop's fourth quarter and official year 2024 earnings conference call. All participants are in listen-only mode in our English line. As a reminder, today's conference call is being recorded. Please note that the discussion today will contain forward-looking statements made under the Safe Harbor Provisions of U.S. Private Securities and the Litigation Reform Act of 1995. Forward-looking statements are subject to risk and uncertainties, and may cause actual results to differ materially from our current expectations. Potential risk and uncertainties include, but are not limited to, those outlined in our public filings with the SEC.

The company does not undertake any obligation to update any forward-looking statements, except as required in applicable law. Also, this call includes discussion of certain non-GAAP narratives. Please refer to our earnings release for our consideration between non-GAAP and GAAP. Joining us today on the call are Mr. Shen Peng, our Founder, Chairman, and CEO, Mr. Ran, Director and GM of Insurance Business, Mrs. Xiaojiao Cui, Head of Finance Department, and Mrs. Jasmine Lee, Board Secretary. We'll be happy to take some of your questions in the Mandarin line at the end of the conference call. Hello, everyone. This is Shen Peng. In 2024, Waterdrop delivers solid performance across all business units, driven by our integrated scenario ecosystem and the sustainable AI investments.

By fully upgrading products and services, we have established new benchmarks for AI implementation in the industry. In Q4, our total revenue was about RMB 687 million. The annual revenue reached RMB 2.77 billion, up by 5.4% worldwide. In terms of the profits and net profits attributable to ordinary shareholders, in this quarter, it was over RMB 99.6 million, and the annual profit result has doubled, achieving our financial guidance to the market. The company has remained profitable for 12 consecutive quarters. I'm also pleased to announce that, based on our stable operations in the second half of 2024 and a significant year-over-year increase in profits with board approval, the company will soon launch its third cash dividend, with $0.002 per ordinary share and $0.02 per ADS.

The shareholders of ordinary shares and ADS, as of the closing date of April 11, 2025, U.S. Standard Time, are entitled to the dividend. Moreover, if our business maintains healthy operations, we plan to distribute dividends semi-annually to reward the support of our shareholders and investors. In addition to dividends, the company also continues to actively repurchase from the open market. As of the end of February 2025, the company has cumulatively repurchased about 52.1 million ADS in the open market, costing approximately $104 million. We remain committed to social responsibility and sustainable development. As of the end of 2024, Waterdrop Welfare Platform has partnered with 115 public charities and hosts over 15,000 charitable projects.

Furthermore, we will also find many ESG achievements in Q4, including a ESG rating from United Equity and Chu Jing. As a milestone, Waterdrop's medical crowdfunding has been officially recognized as a pioneer in China's medical crowdfunding sector. Following new legislation changes, the platform secured designation from the Ministry of Civil Affairs on December 25, 2024, becoming one of the first batch state-endorsed online service platforms for individuals seeking financial help. Since China's Spring Festival, DeepSeek has sparked global attention in AI innovation. We believe the AI-powered industry will redefine productivity. As a tech-driven company, we have launched AI tools and agents using large language models in 2024, especially in online sales and broker support. With DeepSeek integration, we are accelerating AIGC development.

At the end of today's earnings call, we will share our insights and latest breakthrough on AI. For 2025, we will focus on business growth, targeting 20% annual revenue growth with quarterly increments. Based on current projections, we expect financial performance of Q1 to maintain stable growth and sustained profitability. Subsequent quarters may see strategic investments aligned with accelerated revenue growth. That concludes my remarks. Now, I will turn it over to Xiaojiao Cui for the financial update. Thank you, Chen Peng. Before I go into details, please be reminded that all numbers quoted here will be in RMB, and please refer to our earnings release for detailed information on our financial performance on both the year-over-year and quarter-over-quarter basis, respectively. In Q4, the company delivered solid performance, with total revenue reaching RMB 687 million, up 4.2% YOY.

For full year 2024, our revenue totaled RMB 2.77 billion. In 2024, our non-insurance business accounted for near 15% of total revenue. Our crowdfunding service fees amounted to RMB 267.7 million, while digital clinical trial solution income reached RMB 91.1 million, collectively reflecting diversified revenue streams. Through AI-driven efficiency enhancement and refined operation management, our total operating costs and expenses in Q4 amounted to about RMB 634 million, remaining stable compared to the same period last year. For the full year of 2024, overall operating costs and expenses decreased by 1.7% from last year. Breaking this down, in Q4, operating costs increased by 7.2% year-over-year to RMB 323.8 million. This was primarily driven by an increase of RMB 39.1 million in cost of referral and service fees, partially offset by a decrease in personnel costs.

Sales and marketing expenses increased by 4.1% year-over-year to RMB 182 million this quarter. This resulted from our strategic expansion in customer acquisition due to an increase of RMB 30.2 million in marketing expenses for third-party traffic channels, and partially offset by a decrease of RMB 22.8 million in personnel costs. General and administrative expenses declined significantly to RMB 73.7 million, down 23.2% year-over-year, mainly attributable to a decrease in allowance for credit loss and professional service fees. R&D expenses stood at RMB 54.3 million, a 9.3% decrease from the same quarter of last year, primarily reflecting optimized personnel costs. However, our AI investment remains unwavering.

In fact, we plan to increase investments in 2025 to further strengthen our technology edge. We have sustained remarkable profitability and doubled our annual net profit attributable to our ordinary shareholders. We continue to generate a positive operating cash flow during this quarter. As of December 31, 2024, our cash position was RMB 3.67 billion, providing robust support for future growth initiatives. This concludes our financial review. I will now pass to our Board Secretary, Jasmine, to update our business development. Thank you, Xiaojiao. In Q4, our insurance business achieved an SIP of about RMB 1.9 billion, representing 24.5% year-over-year growth, while showing a 7% quarter-on-quarter decrease. Insurance-related revenue reached RMB 580 million, down 1% year-over-year, and 3% quarter-on-quarter, while operating profits stood at RMB 103 million, with an 18% operating margin.

During this quarter, our short-term insurance SIP surged 44.2% YOY to RMB 1.43 billion, driven by breakthroughs in public customer acquisition through the real-time marketing optimization and the new payment channels. Leveraging user insights, we introduced more inclusive short-term insurance products in this quarter, including medical plans tailored for individuals with pre-existing conditions. For long-term insurance products, SIP reached RMB 460 million, with a slight sequentially decreased impact by the product suspension of last quarter and the holiday reason in this quarter. In Q4, we optimized our team staffing while maintaining high productivity per capita. Furthermore, we sustained our efforts to enhance and diversify our long-term insurance offerings with disability income insurance, along with achieving an SIP exceeding RMB 90 million.

The company is also continuing to develop its online planner model. In the fourth quarter, Shenzhen Bao independently contributed approximately RMB 90 million in SIP. For our medical crowdfunding business, as of the end of 2024, around 470 million people accumulated. They donated RMB 67.5 billion to about 3.4 million patients through our platform. Our platform processed thousands of campaign applications and over 50,000 image submissions, with risk control remaining a top priority. To enhance efficiency, we implemented a DUE model AI framework. On one hand, lightweight models now detect pixel-level image tampering in milliseconds, reducing false positive rates from 0.3% to an industry-leading 0.1%.

On the other hand, summarizing experience from the accumulated cases from the past eight years, extracting more than 1,700 characteristics to assist medical consultants in making reasonable predictions of the diseases and reducing the risk of cost by fraud and misappropriation. Following 2024's regulation updates and government official designations, Waterdrop, as China's largest online service platform for individuals seeking financial help, will continue setting standards through technology-driven transparency and user-centric operations. During this quarter, our digital clinical trial solution income continued its growth, realizing a year-over-year increase of 22.9%. We collaborated with a total of 177 pharmaceuticals and CROs. Moreover, we enrolled 880 patients and initiated service for 79 new programs during the fourth quarter of 2024.

In addition, the platform leveraged AI models to enhance doctor-patient service across areas like health management, automatic science content creation, and education digital avatars. Multiple initiatives have been launched, boosting efficiency and user experience while ensuring compliance. This is a key progress of each business line in this quarter. Next, we will invite Ran Wei, the GM of the insurance business, to deliver a keynote on Waterdrop's AI-driven insurtech. Thank you, Ju. Dear investor and analysts, it's a great honor to present Waterdrop's achievements in AI and other insurtechs during the earnings conference. Waterdrop has been deploying AI since 2019, starting with machine learning algorithms, OCR, and NLP technology to optimize insurance product recommendations, risk matching, and risk control, boosting online conversion rates and service efficiency.

We established a series of digital infrastructures, including data middle platforms, intelligence marketing systems, smart CRM systems, and AI risk control systems, enhancing user experience while strengthening our insurtech leadership. This journey also cultivated a cross-functional R&D team, combining insurance expertise with AI capabilities. Amid this transformative wave, Waterdrop has developed insurtech competitiveness barriers for AI applications, which can be divided into four core areas. Firstly, we have built a multi-dimensional ecosystem, interacting text and voice interactions, digital avatars, and other innovations, covering channels such as WeChat, WeCom, and AI outbound calls. This enables seamless transaction between services while accumulating multi-dimensional data.

Secondly, we have established a fully digitalized service chain, spanning demand identification, product matching, underwriting, and claims processing. Unlike the conversational ability of general-purpose LLMs, our service chain requires not only a massive knowledge base but also precise control over the entire service workflows to ensure high reliability. Currently, we handle tens of thousands of highly complex insurance inquiries in databases. Nearly over eight years, we processed 100 million scale private data assets in the insurance industry, including 112 million operation data on a knowledge base of over 7,000 products and massive structured service data entries. This real-world data is irreplaceable training material for building insurance-specific LLMs.

Lastly, distinct from general-purpose LLMs, Waterdrop has developed a vertical knowledge graph for insurance scenarios, combining recommendations, insights, and service conversions. Across hundreds of refined operational nodes, we employ specialized tools such as family coverage gap analysis algorithms, customer design decision-making psychology models, and resolution frameworks for customer objections. For example, our model can generate a customized insurance plan in real time based on user blogging duration, consultation keywords, family structure, and other features. This four-in-one capability system integrates scenarios, service data, and knowledge, ensuring the rapid implementation of Waterdrop's AI solutions. While general-purpose LLMs still struggle with hallucinations, our vertical models can already deliver definite answers to our users.

Overall, the combination of the multiple scenarios and online data has offered irreplaceable strategic value in insurance LLMs. Next, I will demonstrate how our AI has delivered memorable results. Over the past two years, we have actively adopted cutting-edge fundamental LLM technologies while developing AI applications tailored to our business scenarios. I want to share some examples. Firstly, our insurance AI agent has end-to-end self-decision-making capability. It integrates the ability of possession, inference, and decision execution, leveraging multi-agent collaboration technology. In the medical insurance scenario, it enables fully automatic decision-making up to 15-minute conversations. For example, a single call can automatically complete a process like understanding user needs, health risk modeling, and the policy interpretation.

Our AI agent can support modular agent units collaboration in real time and the flexible transition across health insurance, auto insurance, and other scenarios. The system spans over 7,000 products and can respond within two seconds. Moreover, our AI agent can dynamically optimize LLM algorithms by learning from the operation feedback and self-outcomes through over 100,000 data interactions. Secondly, we also continue to optimize our model through two main approaches. The first one is knowledge distillation. This method uses technology like SFT and chain of thought reasoning. It impacts millions of insurance-related knowledge to the model, making it comparable to the senior human component. The second is reinforced fine-tuning of the sales strategies. We integrated the DeepSeek and other open-source frameworks with our self-developed GRPO algorithm.

This has increased model accuracy by 20%. Furthermore, our AI agents can make multi-modal interactions with our users, such as real-time coordination of voice, text, and the image, minimizing user waiting time. We also upgraded it with an emotion-adaptive system. It can dynamically adjust its speaking speed and modify its tone based on semantic analysis. For example, our AI agent will slow down by 20% when it interacts with elderly users. Over the past two years, our 48 LLM-related patents have been granted, with over 100 patents filed and nearly 100 AI innovations. Waterdrop is one of the first insurtech companies capable of enabling AI agents to independently close deals in complex scenarios. Next, I want to show you our detailed AI application results.

Following a development path from simple automation to advanced integration, our LLM application now focuses on two directions: co-pilot and AI agent directing and serving users. For our co-pilot application, we have quality control co-pilot and self-co-pilot. Our quality control co-pilot can replace the manual sampling of service recordings and perform text analysis. With an 88% accuracy level of human voice across voice and text channels, we expect the frontline quality control cost to be reduced by 40%, significantly improving efficiency. Our self-co-pilot can enhance self-efficiency by enabling rapid summarization of client intent and tailored product recommendations.

This tool elevates the service quality and conversion rates, empowering our human agent to better meet client needs and achieve the self-targets. About our AI agent, we have three different applications, including AI Health Insurance Agent, AI Service Assistant, and AI Insurance Planner. For our AI Health Insurance Agent, it can directly serve customers via voice-based consultations, like single sessions lasting up to 15 minutes, outperforming junior agents in efficiency. It currently generates millions of insurance premiums per month, projected to reach tens of millions by year-end. Building on this, our AI auto insurance agent has entered data training with two P&C insurers, with selected projects already in user beta testing. For our AI Service Assistant, Xiaojiao Cui, embedded in our app, this 24/7 AI assistant handles Q&A about user policy and the products you are interested in.

By 2025, we expect AI to reduce case transferring rates by 50% and improve efficiency by 50% compared to human agents. Leveraging DeepSeek's long-chain reasoning ability, we are developing an AI insurance planner to provide real-time, high-quality family insurance consultation and planning services, enhancing user experience and growth potential. Talking about the future outlook, in the next three to five years, along with the AI technology and the application trends that I just mentioned, based on our industry know-how, we will further consolidate our leading position in the insurtech industry. Thank you for listening. Ladies and gentlemen, with that, we will conclude today's earnings conference call. We do thank you for joining. Have a good time.