We provide intelligent marketing services to enterprises by leveraging AI technologies through our proprietary AI application products. During the Track Record Period, we delivered intelligent advertising services and intelligent data management through our two flagship platforms, AlphaDesk and AlphaData, respectively.
We were among the first companies in China to apply AI technologies to enterprise digital transformation in marketing and sales. By integrating AI technologies with business scenarios, we have developed a product foundation based on technology, data intelligence, and industry expertise. In addition to our established flagship platforms, AlphaDesk and AlphaData, we have launched Deep Agent in 2025, a series of enterprise AI agent products that leverage open-source large language models (“LLMs”) to further enhance marketing automation and efficiency.
Both AlphaDesk and AlphaData were independently developed by us. Except for commonly used development and programming tools, database products and algorithmic technologies, such as Java, Python and JavaScript, we do not rely on any other third-party technologies. The embedded AI models and system architecture are proprietary and do not rely on open-source AI models.
Our Business Model
During the Track Record Period, we generated our revenue from the provision of (i) intelligent advertising services, and (ii) intelligent data management.
Intelligent Advertising Services
Our intelligent advertising services operates through AlphaDesk, a platform designed to optimize advertising performance through real-time decision-making and automated execution.
Prior to the commencement of each campaign, advertisers set their specific campaign plans, which typically include campaign duration, total budget, and performance objectives (e.g. target audience, region, media platforms, duration, target click-through volume or cost per conversion). Advertisers are also required to provide ad creatives, as well as third-party tracking links for monitoring impressions and clicks.
Based on the campaign plans set by advertisers, AlphaDesk leverages machine learning models to predict the likelihood that an end-user will click on or convert through a particular advertisement. These models are trained using large volumes of historical advertising data, including past impression records, end-user interaction behaviors, and campaign outcomes.
An advertising impression opportunity refers to a biddable chance provided by a media platform to display an ad to a specific end-user. Each advertising impression opportunity is tied to a particular end user and only becomes available when that user is actively browsing the media platform. Upon such occurrence, the media platform transmits relevant data to AlphaDesk in real time. AlphaDesk’s machine learning models then evaluate this opportunity by analyzing features, such as the end user’s historical engagement patterns, the characteristics of the ad slot (e.g. format, placement, and size), and historical campaign performance, to predict the likelihood of end user engagement. Based on this prediction, AlphaDesk determines the optimal bid price to meet the advertiser’s KPIs (e.g. target cost-per-click or return on ad spend), and automatically submits a bid. Once the bid is won, AlphaDesk also selects the most suitable creative for display, all within milliseconds.
AlphaDesk is provided as a cloud-based platform and does not require installation by customers. Customers access AlphaDesk online pursuant to the terms of their service agreements with us. Upon expiry or termination of such agreements, customers cannot continue to access or use the AlphaDesk platform.
During the Track Record Period, we worked with a number of leading media platforms in China, including but not limited to Douyin, Tencent, and Kuaishou. The majority of our contractual arrangements with these platforms were entered into through designated resellers, which are authorized partners of the relevant platforms. We also maintained direct cooperation with certain platforms. AlphaDesk connects directly with media platforms for campaign execution and coordination, regardless of whether the contractual arrangement is made through resellers or directly with the platforms.
We charge fees based on the agreed KPIs and pricing terms for each campaign, rather than a fixed subscription fee. We recognize revenue upon completion of each campaign. Under our intelligent advertising services, we adopt two revenue models based on the scope of services provided to customers. In most cases, we act as the principal and manage the full execution of advertising campaigns. Under this model, we manage the entire campaign lifecycle, including media resource procurement, campaign execution, and post-campaign performance analysis. We also assume responsibility for achieving the customer’s performance KPIs. We charge customers a pre-agreed advertising budget, and recognize the full amount as revenue. All related delivery costs, including media resources acquisition costs, are borne by us.
In other cases, customers independently procure media resources from media platforms or resellers and use our proprietary AlphaDesk platform to manage their campaign execution. For these engagements, we charge a platform usage fee, which we recognize as revenue on a net basis.
Customers typically enter into a master service agreement with us, and fees are settled following the completion of each campaign. Platform usage fees are either (i) calculated as a percentage of the advertising spending executed through AlphaDesk, or (ii) charged at a fixed fee as specified in the service agreement. The applicable pricing model is determined through commercial negotiation with the customer.
Although AlphaDesk’s core capability lies in real-time bidding across media platforms, certain customers may prefer to purchase media resources directly from specific platforms or resellers. In such cases, while customers may have access to designated media resources, ads can only be served in real time when advertising impression opportunities arise during end-user activity. AlphaDesk remains integral to these campaigns by enabling real-time execution and optimization.
For example, AlphaDesk can control the number of times the same advertisement is shown to an individual end-user across different media platforms, thereby minimizing ad fatigue and improving campaign efficiency.
The key difference between the two models is the scope of our service responsibilities. Under the full execution model, we deliver end-to-end campaign services, covering media resources procurement and performance delivery, and take responsibility for KPIs. Under the campaign management model, the customer keeps responsibility for media resources procurement and performance outcomes. We only provide access to AlphaDesk to support campaign delivery and optimization.
We primarily operate AlphaDesk to deliver intelligent advertising services to our customers. In most cases, our team configures and manages campaign execution for customers through AlphaDesk. In a limited number of situations, we also grant customers direct access to AlphaDesk accounts to manage their own campaigns. Our team then offers backend support and optimization services as needed.
Intelligent Data Management
The core product of our intelligent data management business is AlphaData, our intelligent platform for CRM. AlphaData enables enterprises to establish a centralized customer data platform built on their proprietary data, thereby breaking down data silos. On the basis of this consolidated data, AlphaData applies AI algorithms to predict key stages of end-user engagement, such as product preferences and repurchase cycles, and to automate customer relationship management.
For example, a well-known automobile brand used AlphaData to improve customer retention and promote vehicle repurchases. AlphaData integrates customers’ historical data, including vehicle profiles, ownership information, and after-sales service behavior. The proprietary repurchase prediction model embedded in AlphaData scores each end-user’s likelihood and timing of repurchase. High-scoring end-users were routed to the customer’s sales team for one-on-one follow-up, while medium-scoring end-users entered a targeted nurturing program. According to customer feedback, the average in-store visit rate of end-users in the highest scoring tier was approximately 2.72 times that of end-users before using AlphaData.
AlphaData’s prediction capabilities are supported by AI models trained on large volumes of historical behavior data. AlphaData tracks key attributes such as user behavior, content engagement, transactions, and demographics. It assigns predictive scores at different stages of the user lifecycle. These scores help automate actions like audience segmentation, message personalization, and timing of outreach to improve efficiency and conversion.
This business includes delivering software to customers and deploying it on the customer’s designated cloud environment, along with implementation services to integrate AlphaData with their internal systems. Such implementation services primarily include interfacing with the customer’s existing systems (such as CRM systems), conducting customized development based on customer requirements, and integrating and processing data in accordance with the customer’s data governance framework and data formats. Our intelligent data management contracts with customers usually cover multiple services, including software licensing, implementation, and ongoing support.
After software delivery and acceptance, customers obtain the right to use the software and enjoy related services as agreed.
AlphaData is deployed on the customer’s designated cloud environment as part of our implementation services, with the customer’s internal data serving as its data source. We offer two types of licenses for AlphaData: perpetual licenses and subscription licenses. Access is controlled through user accounts, with built-in expiration settings linked to the applicable license terms. For customers who acquire a perpetual license through a one-off purchase, they are entitled to use the purchased version of AlphaData on an ongoing basis. However, upgrades, maintenance services, and newly released versions are available for an additional fee. For customers who acquire a subscription license, the right to use AlphaData is granted only for the subscription period. Upon expiration or termination of the subscription, customers must renew the subscription to continue accessing AlphaData; otherwise, access will be automatically suspended and further use will not be permitted.
During the Track Record Period, the terms of these contracts ranged from one month to five years. We typically charge fees based on the agreed service schedule, with billing cycles generally within one year. Upon contract expiry, customers may choose to renew or purchase additional services based on their needs. We generate revenue through (i) standardized platform licensing fees on a one-off or annual basis subject to our negotiation with customers, (ii) implementation and deployment service fees, and (iii) fees for platform upgrades and ongoing maintenance.
Development of Our AI Application Product
In February 2025, we launched Deep Agent, our enterprise AI agent system designed to address various marketing and sales scenarios. The system comprises a series of agent products that leverage agentic technologies and domain-specific know-how to perform tasks, such as AI-driven analytics and consumer insights generation. Leveraging open-source LLMs and powered by domain-specific machine learning models, it enables enterprises to automate operational processes and enhance decision-making. Customers may subscribe to individual agents within the Deep Agent system based on their business needs. Although Deep Agent had not generated material revenue up to the Latest Practicable Date, we had secured 33 contracts for its provision as of the Latest Practicable Date, with an aggregate contract value of approximately RMB22.9 million.
Our Customer Base and Brand Recognition
Since inception, we have cultivated a customer base with a strategic focus on medium-to-large enterprises with complex decision-making needs and advanced levels of digitalization. During the Track Record Period, we served approximately 468 end customers across various industries, including 69 Fortune Global 500 companies. Our end customers operate in sectors including e-commerce, fast-moving consumer goods (“FMCG”), automotive, retail, beauty, and hospitality.
Our scenario-based, results-driven solutions have enabled us to maintain customer loyalty, as reflected by high net dollar retention rates across our core products. During the Track Record Period, AlphaDesk achieved a net dollar retention rate of over 85%, and AlphaData maintained a rate of over 80%.
Our brand, “Deepzero,” has been widely recognized within the industry, supported by our designation as the National-Level “Little Giant” Enterprise and acknowledgments from global research firms for our performance in key AI application categories.
Source: DeepZero Tech (02723) Prospectus (IPO Date : 2026/05/18) |