Run ad-hoc SQL queries against ingested datasets

This task produces query results that answer a specific analytical question using data already ingested into the platform — without requiring data export, ETL to a separate analytics store, or proprietary query syntax. The output may be a result set displayed in a terminal or query editor, a downloaded CSV, or a visualization in a connected BI tool. Ad-hoc SQL capability is the primary analytical surface for data engineers and analysts who need to validate ingestion completeness, explore data distributions, or answer business questions directly from platform data.

XDM dot-notation addressing. Unlike relational databases where all columns are flat strings at the table level, XDM datasets store complex object types. Querying them requires dot-notation paths that mirror the XDM schema tree: placecontext.geo.city, commerce.productViews, productListItems.name. The path to any field can be found in the Schema Editor's field-detail panel, which provides a copy-path button. Selecting a complex object without specifying a leaf field (e.g., SELECT placecontext.geo FROM ...) returns the entire object serialized as a string — usually not the intended behavior.

Query timeout and CTAS. The Query Service default timeout is 5 minutes (300 seconds). Queries involving large datasets with complex joins or aggregations may exceed this limit. The recommended pattern for long-running analytical queries is CTAS (Create Table As Select), which executes the query asynchronously and writes results to a named dataset that can be queried later without timeout risk. The new dataset is immediately queryable from any connected BI tool.

BI tool connectivity. Any application that supports the PostgreSQL protocol can connect directly to AEP Query Service using the PSQL connection string displayed in the Query Service UI. This includes Tableau, Microsoft Power BI, Looker, Qlik, and others. The connection string includes a host, port, database name, user, and password; no additional drivers are required beyond the standard PostgreSQL JDBC/ODBC driver.

Parallel viability (high). SQL-based ad-hoc analytics against ingested data is available in every CDW-native CDP: Snowflake, BigQuery, Databricks, Redshift. The XDM dot-notation addressing is AEP-specific; equivalent platforms use flat column names or semi-structured JSON path syntax (e.g., data:event_properties:product_name in BigQuery). The analytical skill — enumerate tables, understand schema structure, write SQL, handle timeouts — transfers directly. Phase 3 will document equivalent Snowflake query patterns.