> For the complete documentation index, see [llms.txt](https://web3-growth-agent-wga.gitbook.io/whitepaper_ver2.0_en/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://web3-growth-agent-wga.gitbook.io/whitepaper_ver2.0_en/.-wga-solution-overview-and-structure2/2.-universal-data-flow.md). # 2. Universal Data Flow WGA data processing consists of five distinct stages: * **Stage 1: Collection** Behavioral events occurring across Web2 and Web3 environments are gathered via SDKs, APIs, and pixels. This includes page views, clicks, registrations, transactions, contract interactions, and community engagement. WGA targets events already occurring within apps, websites, and on-chain. When a partner application integrates the WGA SDK, the resulting user behavioral data is transmitted to the WGA network. * **Stage 2: Normalization** Disparate event formats, timelines, and definitions across various channels are standardized. Events with identical functional meanings, such as "Purchase Complete" on one platform and "Conversion" on another, are normalized into a single standard event structure.. * **Stage 3: AI Validation** Normalized data passes through an AI validation pipeline where four concurrent analyses are performed: * Fraud Detection: Isolation of bots, automation, redundant accounts, and anomalous patterns. * Sentiment Analysis: Interpretation of interest intensity and response levels within the behavior. * Pattern Recognition: Identification of recurrence, engagement depth, and attrition segments. * Network Analysis: Tracking of acquisition paths, diffusion structures, and connectivity. These analyses apply varying weights and criteria based on behavioral context, channel characteristics, and campaign types. Results are expressed as confidence scores rather than binary "valid/invalid" labels. * **Stage 4: Metric Structuring** Validated data is converted into the proprietary WGA KPI framework. These KPIs are designed for long-term comparability, integration with external datasets, and reusability within the analytical market. * **Stage 5: Settlement and Delivery** Interpretative results and core summaries are anchored to the blockchain. The recorded data is not raw logs, but a summary of the validation criteria, the interpretive process, and the resulting KPI values. This record is immutable and allows for transparent auditing of the verification process. Final outputs are delivered via reports, dashboards, and APIs.