BigQuery augmented analytics: investigating metrics in SQL
Analytical workflows that previously required separate statistical notebooks or specialised tooling are increasingly becoming callable warehouse functions.
What happened
Google’s 14 September announcement describes six BigQuery table-valued functions for key-driver analysis, causal-effect estimation, correlation, change points, trends and seasonality. They produce structured results within warehouse workflows. The announcement groups capabilities together; availability must be checked per function. For example, the release notes label AI.CAUSAL_EFFECT as Preview on 10 September.
Why it matters
BI traditionally answers what happened. Diagnosing why usually requires additional investigation. Warehouse-native diagnostic functions reduce the technical distance between detecting an unusual KPI and investigating its drivers.
Who it affects: BI developers, analysts, data scientists, analytics engineers and teams building conversational analytics.
What to do next
Identify recurring analytical questions such as: • Why did this KPI change? • When did the underlying behaviour shift? • Is this movement seasonal? • Which variables appear most associated with it? Test whether warehouse-native analytical functions can standardise those investigations.
Signal Take
SQL is becoming an interface to statistical investigation as well as aggregation. Keeping the work near governed data can improve reproducibility, provided teams preserve assumptions and distinguish correlation from causal evidence.
Scope and limitations
Statistical associations are not automatically causal explanations. Check each function’s status, assumptions and required inputs before using its output in decisions.
Sources and editorial record
Google Cloud — augmented analytics (opens in a new tab)- Source type
- Vendor documentation or announcement
- Source published
- 14 Sept 2026
- Source checked
- 19 Sept 2026
Prepared with AI assistance and checked against the linked source. This is editorial interpretation, not an independent product benchmark. How we work.