twitter

website

github

discord

Blockchains (L3)

# XAI Active users (weekly) Methodology

## 1. Overview

XAI Weekly Active Users (WAU) measures the number of unique addresses that initiated at least one transaction on the XAI blockchain within a 7-day rolling window. This metric provides insight into weekly user engagement and adoption trends for the XAI Layer 3 blockchain. The calculation uses HyperLogLog algorithms for efficient approximate cardinality counting and generates multiple dimensional breakdowns to support various analytical perspectives.

## 2. Data sources

- **XAI blockchain transactions**: Raw transaction records from the XAI Layer 3 blockchain, including transaction hashes, sender addresses, recipient addresses, timestamps, and gas usage data
- **Transaction metadata**: Gas prices, gas used amounts, and block timestamps for each transaction
- **Internal processing metadata**: Table modification timestamps used for incremental data processing and ensuring data freshness

## 3. Step-by-step calculation

### Component A: Transaction data extraction
- **Purpose**: Extract and standardize XAI blockchain transaction data for user identification
- **Logic**: The system retrieves all transactions from XAI blockchain records, filtering for transactions up to August 22, 2025. Each transaction is enriched with standardized metadata including the XAI token address (`0x82af49447d8a07e3bd95bd0d56f35241523fbab1`), chain identifier, and market classification. Gas fees are calculated by multiplying gas used by gas price and converting from wei to ether units.
- **Key fields**: `block_timestamp`, `transaction_hash`, `from_address`, `to_address`, `gas_price`, `gas_used_amount`, and calculated `l3_gas_fee_amount`

### Component B: Daily active user identification
- **Purpose**: Identify unique user addresses per day from transaction activity
- **Logic**: Transaction sender addresses (`from_address`) are extracted and aggregated daily to identify active users. Each unique sender address on a given day is classified as a daily active user with the type "Transaction sender". Timestamps are truncated to daily granularity to ensure proper daily grouping. The system applies incremental processing logic to only process recent data when updating existing results.
- **Key fields**: `timestamp` (daily), `chain`, `user_type`, `user_dau` (the unique address)

### Component C: HyperLogLog sketch creation
- **Purpose**: Convert daily user sets into probabilistic data structures for efficient aggregation
- **Logic**: Daily active user sets are converted into HyperLogLog sketches using 15-bit precision. This probabilistic data structure allows for efficient approximate counting of unique users across time windows without storing individual addresses. Each daily sketch represents the set of unique users for that specific day.
- **Key fields**: `hll` (HyperLogLog sketch), `timestamp`, `chain_id`, `user_type`, `product_id`

### Component D: Datespine and rolling window generation
- **Purpose**: Create complete time series and 7-day rolling windows for WAU calculation
- **Logic**: A datespine is generated covering the full date range from the earliest available data to the current date, ensuring no missing days in the final time series. For each day in the datespine, a 7-day window is created by generating timestamps for that day plus the previous 6 days. This creates overlapping 7-day windows that enable rolling WAU calculations.
- **Key fields**: `timestamp` (target day), `joinday` (days within the 7-day window)

### Component E: Weekly user aggregation
- **Purpose**: Merge daily HyperLogLog sketches across 7-day windows to calculate unique weekly users
- **Logic**: The system joins the rolling window structure with daily HyperLogLog sketches, matching each window day with its corresponding daily user sketch. HyperLogLog sketches are then merged across all 7 days in each window using the `hll_count.merge_partial` function, and the final unique count is extracted using `hll_count.extract`. This produces the approximate count of unique users who were active at least once during each 7-day period.
- **Key fields**: Merged `hll` sketches and extracted `user_wau` counts

### Component F: Dimensional breakdowns
- **Purpose**: Create multiple analytical views of WAU data across different dimensions
- **Logic**: The weekly user data is aggregated into four distinct dimensional views: project-level (all chains combined), chain-level (by individual blockchain), user-type-level (by transaction sender classification), and product-level (by specific product identifier). Each breakdown nullifies irrelevant dimensions while preserving the relevant grouping dimension. All breakdowns are combined using union operations to create a comprehensive dimensional dataset.
- **Key fields**: `dimensions` (breakdown type), `chain_id`, `user_type`, `product_id`, `user_wau`

### Final aggregation
- **Combining components**: The dimensional breakdowns are combined through union operations, creating a unified dataset with multiple analytical perspectives of the same underlying WAU metric
- **Schema standardization**: The `finalize_metric_user_wau` macro applies standard schema formatting, ensures consistent field types and names, and implements the "fill_zeros" strategy to populate missing periods with zero values
- **Time boundaries**: Data is filtered to exclude periods beyond August 22, 2025, and incremental processing ensures only recent data is reprocessed during updates
- **Final calculation**: The final WAU value represents the approximate count of unique addresses that initiated at least one transaction within any 7-day rolling window, calculated using HyperLogLog cardinality estimation for computational efficiency

*This methodology was automatically generated by Token Terminal's methodology analyzer.*

Unique revenue-generating users over a rolling 7-day window.

XAIXAI

Archived

Active users (weekly)

Active users (weekly)

Competitive landscape

Chain breakdown