Sansan, Inc.

Thank you for joining our earnings presentation today. I am Hashimoto, the CFO of Sansan, Inc. I will present our quarter three results for the fiscal year ending May 31, 2026, fiscal year 2025. Please turn to page four. Here are the highlights of our quarter three year to date of fiscal year 2025. First, net sales increased by 26.1% year over year, maintaining steady growth. Adjusted operating profit rose by 131.1% year over year, reaching a record high, driven in part by reduction in losses from our accounting, AX Solution Bill 1. Second, Bill 1 continued its high sales growth, with a net increase in MRR expanding for four consecutive quarters. Third, in light of our solid performance through Quarter 3, we have revised upward our full-year earnings forecast for fiscal year 2025 and medium-term financial policy fiscal year 2026. I will move on to the results for the three months of Quarter 3. Net sales grew steadily, up 25.3% year-over-year, while gross profit margin improved by 0.4 percentage points year-over-year. Adjusted operating profit rose sharply by 69.6% year-over-year, driven by both sales expansion and a decline in the SG&A ratio, marking a record high for a single quarter. Against this backdrop, both ordinary profit and profit attributable to owners of parents saw significant increases. I will now explain the details of adjusted operating profit. Net sales increased steadily by 25.3% year-over-year, and the cost of sales ratio decreased by 0.4 percentage points year-over-year. Advertising expenses increased by 51.1% year-over-year due to intensified promotional activities, such as television commercials, driven by strong order intake for Bill 1, and the ratio of advertising expenses to net sales rose by 1.7 percentage points. Meanwhile, the increase in personnel expenses was limited to 11.7% year over year, and the ratio of personnel expenses to net sales fell by 4.0 percentage points. As a result, the S, G, and A ratio decreased by 6.1 percentage points year-over-year, and while continuing to make the investments necessary for growth, adjusted operating profits saw a significant increase of 69.6% year-over-year. Next, I will talk about the overview by segment. the Sanson and Bill One business continued to report steady net sales growth and a significant increase in adjusted operating profit. Although the Eight business continued to report strong net sales growth, it reported a decrease in adjusted operating profit 24.3% year over year. Nevertheless, it remained profitable. Next, starting on page eight, I will explain the details by segment. Net sales for the Sansan and Bill One business increased by 23.6% year over year, driven by steady growth in Sansan and continued high growth in Bill One. In addition, others, which includes the performance of Contract One and Nine Out Inc. Cars, a group company, also performed well and contributed to sales growth. of which Contract 1 sales increased by 101.9% year-over-year, and the number of subscriptions reached 653, a 102.2% increase year-over-year. Furthermore, annual recurring revenue has been expanding steadily, exceeding 1.0 billion yen as of the end of March 2026. Adjusted operating profit rose sharply, up 76.8% year over year. By solution, improved profitability in Bill 1 contributed significantly. It turned a profit in a single month in certain months during Q3, and the net loss was reduced to approximately 60 million yen, improving about 800 million yen year over year. Additionally, Sansan reported a 19.2% year-over-year increase in adjusted operating profit. Here is an explanation of KPIs for Sansan. The left side of the slide shows recurring sales with a solid 15.6% year-over-year growth. As shown in the graph in the center of the slide, the number of SanSan subscriptions increased by 12.8% year-over-year on a period-end basis, with a net quarterly increase of 372 subscriptions, accelerating growth from quarter two. the monthly recurring sales per subscription rose by 1.4% year-over-year and showing a high-level performance. The graph on the right side of the slide shows the last 12 months' average of monthly churn rate. In Q3, it stood at 0.54%, which is roughly the same level as in Q2, maintaining at a stable, low rate below 1%. Furthermore, the reasons for cancellation remain unchanged and are based on circumstances specific to each company. For example, we are not aware of any cancellations resulting from companies developing in-house systems to replace SanSan by using generative AI. Next, please turn to page 10 for the KPIs of Bill 1. On the left side of the slide, MRR reached approximately 1,156 million yen, up 36.8% year over year. This represents a net increase of approximately 86 million yen from the end of quarter two, marking the fourth consecutive quarter of growth in quarterly net increase. Next, in the center of the slide, the number of paid subscriptions increased by 36.1% year over year, with a net quarterly increase of 325 subscriptions. This also continues the stable growth of exceeding 300 subscriptions per quarter since Q3 of the previous fiscal year. Both monthly recurring sales per paid subscription and the last 12 months average of monthly churn rate improved slightly year over year and remained at healthy levels. Next, I will explain the development status of AI-powered features for SanSan, BillOne, and ContractOne, as well as how these are expected to contribute to our future performance. First, let me introduce the Sansan MCP server, which we announced in our financial results for fiscal year 2024. This feature enables users to directly retrieve business card-based contact data and contact history stored in Sansan from generative AI tools, such as Microsoft Copilot, that they use within their own organizations. In essence, it serves as a bridge connecting Sansan with external generative AI. Since launching proof of concept last November, we have conducted extensive testing with more than 10 early adopters, primarily large companies, and have identified a key challenge. Specifically, we found that simply feeding data from Sansan to external AI system results in variations in the quality of the output. This is because the accuracy of the responses and the way they are summarized depend on the performance of the AI models used by each company and the instructions provided by users. To address this issue, we are currently updating our system at SanSan to incorporate AI so that data is organized and processed into the optimal format for each specific use case before being passed on to external AI systems. This will enable SanSan to generate highly accurate reports covering, for example, recent news, past interactions, and summaries of activities, which external AI systems can then utilize directly. As a result, users will be able to easily elicit accurate responses from generative AI without having to deal with complex settings or advanced prompt design. Launched last November, Sansan AI Agent differs from the Sansan MCP server mentioned earlier. It is an AI feature that operates entirely within Sansan. With Sansan AI Agent, users can use natural language to access and utilize data within Sansan, as well as data from sales tools integrated with Sansan, all directly within the platform. This Sansan AI Agent was a fairly large-scale feature, essentially a fully customized solution. However, we were able to confirm a clear customer need for it, as evidenced by our success in securing several major contracts. That said, if we were to keep this fully customized version as is, the number of companies able to adopt it would inevitably be limited. Therefore, in order to attract more users going forward, we have decided to reorganize the key features of SanSan AI Agent into SanSan AI Search and make it easier to use within the SanSan platform. For example, using SanSan AI Search, users can instantly generate highly accurate lists by cross-referencing business cards, contact history, and relevant publicly available online information. So, it can even respond to highly abstract instructions such as identifying companies with which we already have a relationship and companies with which we have no relationship among those in the manufacturing sector that are advancing digital transformation. We plan to roll out these features in phases with a view to fully implementing them around the summer We believe that expanding our AI capabilities will not only help us acquire and retain new customers, but also increase unit price per existing customer. We believe that as we develop these features and increase the utilization of the data accumulated in SanSan, the value of the database itself will also grow. The source of this value lies in primary information, such as business cards and contact histories. that generative AI cannot obtain from publicly available information. Because Sansan is structured to continuously collect and accumulate this primary information through the use of its services, we expect that as AI adoption advances, this structural strength will be further reinforced, leading to increased competitive advantage and enhanced value with the generative AI evolution. Next, I would like to explain the current status of AI-powered features in Bill 1 and their contribution to our performance. First, in November of last year, we launched AI Auto-Matching as an optional feature. This feature matches the details of invoices and delivery notes received by a company with the details of purchase-related documents, such as purchase orders and acceptance certificates, at both the total amount and line item levels. Companies conduct these reconciliation processes to prevent overpayments and other issues, but in many cases this work is performed manually by staff, resulting in a significant operational burden. This is an area where both the workload and costs are particularly high, especially in cases where statements contain thousands of lines or where companies must allocate dozens of staff at the end and beginning of each month. Under such circumstances, there is a very high demand for this feature, making it a key priority for improving operational efficiency within companies. Since this feature enables significant reductions in man-hours and costs, We have received a great deal of positive feedback since its announcement. Even before the feature was officially launched, we secured a deal worth approximately ¥42 million in ARR from a single client that included the auto-matching option. Furthermore, in Q3, when the feature was launched, we secured another deal worth approximately ¥28 million in ARR from a single client, indicating a strong start. We anticipate that, as this feature becomes more widely adopted, it will not only increase the average revenue per user for Bill 1, but also accelerate the adoption of Bill 1 invoice receipt and lead to the acquisition of high unit price projects, including those from enterprise clients. Next, let me touch on our AI Auto Request feature, which is scheduled to launch around summer 2026. First, in accounting, voucher preparation work refers to the process of entering accounting items, amounts, and other details based on invoice information to generate vouchers for internal approval. AI Auto Request is a feature that automates the voucher preparation work by using AI to automatically input and suggest accounting items, tax rates, and other details based on the line item data from received invoices and historical invoice data. As staff members create vouchers daily in Bill 1, processing information is accumulated in the learning database, automatically improving the AI's processing accuracy and significantly reducing the workload. In addition, we are exploring a new feature called Automatic Approval, which uses AI to automatically verify the details of submitted vouchers against the invoice data, thereby streamlining the approval process. Although this feature has not yet been released, we have already received a great deal of positive feedback, including new orders placed in Quarter 3 based on the expectation that this feature will be available. Therefore, we expect it to contribute to the future growth of Bill 1. The ability to offer features such as AI auto-matching, AI auto-request, and automatic approval is made possible by the fact that Bill 1 is designed to capture primary information, specifically invoices, and digitize it with high precision down to the line item level. Invoice processing is a process comprising multiple, closely linked stages. ranging from receipt to matching, voucher preparation, approval, and journal entry. To truly achieve business automation through AI, both accurately digitized primary information and a system that consistently supports the entire business process are essential. By capturing the process from the very start of invoice processing, Bill 1 offers the advantage of being able to automate subsequent business processes as a unified whole, which also creates opportunities to expand the value it provides. Finally, I would like to explain the current status of development regarding AI-powered features in Contract 1. While there are currently AI-powered contract review services available on the market, our understanding is that most of them primarily focus on verifying compliance with laws and general review criteria, or comparing contracts against the company's own templates. On the other hand, what is required in the business world are practical decisions that take into account relationships with business partners and past contract terms. To meet these needs, Contract One is launching an AI review feature that enables users to make judgments based on internal standards using historical contract data. Contract One leverages the data digitization and data integration technology developed through SanSan to accurately identify and link contract information for each counterparty ranging from past contracts to the latest memorandums of understanding and organize it into a structured database. This enables users to utilize generative AI to review contracts based on their company's unique criteria, such as assessing whether terms are favorable or unfavorable, while taking into account past similar cases and the contractual history with the relevant party. In terms of implementation, we will build an environment that allows users to access the ContractOne database via the MCP server from the generative AI tools they normally use, such as Claude. As a result, users will be able to obtain responses based on past contractual decisions and review contracts according to the terms specific to each counterparty, all from within their familiar AI environment. Contract One AI Review is scheduled to launch in April 2026. By further enhancing the value proposition of Contract One, we aim to drive further business growth, including the acquisition of new contracts. I will now explain the eight business on page 16. Net sales from B2C services increased 9.9% year over year. Net sales from B2B services increased by 43.8% year over year, driven by steady growth in business events and strong growth of recruiting platform. As a result, net sales for the eight business increased by 41.1% year over year. Adjusted operating profit decreased by 24.3% year over year due to a slight decline in profitability resulting from our efforts to commence new large-scale events aimed at sustaining robust growth in the coming fiscal year and beyond. However, we remain in the black. We have achieved a significant increase in profit for the quarter three year to date, so there are absolutely no concerns. Next, let me explain the upward revision to our earnings forecast on page 18. In light of our performance through quarter three, we have revised upward our full-year earnings forecast for fiscal year 2025 and our medium-term financial policy, so I will explain these in order. First, please turn to page 19 for details on the upward revision to our forecast for fiscal year 2025. Against the backdrop of steady growth across all our solutions, we have raised the lower end of our net sales range and now expect a year-over-year increase of 25.0%, up from the previous forecast of 24.0%. In addition, we have raised the lower end of the range for adjusted operating profit, projecting a year-over-year increase of 143.0%, up from 126.0%. and expect the adjusted operating profit margin to be 16.0%, up from 15.0%. Next, please turn to page 20 for our medium-term financial policy. Previously, our medium-term financial policy set the target range for next fiscal year's adjusted operating profit margin at 18% to 23%. However, in light of the strong profit performance this fiscal year, we are raising the lower limit and revising the target range upward to 20% to 23%. Please note that at this time, we have only revised the adjusted operating profit margin outlook. Details regarding net sales, adjusted operating profit, and other metrics will be disclosed during the full-year financial results announcement planned for July 2026. Looking back at our performance to date, both net sales and profits have been trending steadily, with profits in particular growing at a pace that exceeds our expectations. This is largely due to the significant reduction in losses at Bill 1, which continues to experience high growth, as well as the fact that Sansen, our major revenue source, has entered a phase where it is consistently generating profits. We recognize that, amid the rapid development of generative AI, there remains a sense of caution regarding the medium- to long-term growth prospects of SaaS companies in the current stock market. As we have explained, we view the advancement of generative AI not as a risk, but as a business opportunity. By building on this strong current performance into the next fiscal year and the year after, we will achieve further sales growth and improved profit margins. That concludes my presentation. Thank you very much.