Google BigQuery Connectors Change Schemas in June 2026 and Your Reports Break

Two June Deadlines That Break Different Things

The GA4 connector retention change on June 1 and the Google Ads connector schema change on June 15 affect different tables in your BigQuery dataset — but both break reporting without warning if you haven’t prepared.

The GA4 BigQuery connector requires a minimum 37-month data retention window by June 1, 2026. This isn’t a new feature — it’s a constraint change on existing Data Transfer configurations. If your GA4 export is configured with a shorter retention period, the connector update may not preserve data outside the new retention window. For WooCommerce stores using year-over-year comparison reports, losing months of historical data means losing the baseline those reports depend on.

The GA4 BigQuery connector requires a minimum 37-month data retention window by June 1, 2026 — tables outside this window risk data loss during the connector update (Google Cloud, 2026).

The Google Ads connector adopts Ads API v23 schema changes on June 15, 2026. This is a structural change — date columns move to new data types, some column names change to match the v23 resource naming, and new fields appear that weren’t in earlier schema versions. Any query, view, or dashboard that references the old column names will break.

The dangerous part: these queries don’t throw errors. They return null values for the changed columns. Your Looker Studio dashboard still loads. The numbers just quietly go wrong.

The GA4 37-Month Retention Trap

GA4’s BigQuery export retention window must be set to at least 37 months before June 1 — stores that configured shorter windows risk losing historical event data they can’t recover.

When Google updates the GA4 Data Transfer connector, it applies the new retention constraint to your existing configuration. If your retention window is set to 14 months — a common default for cost management — the connector update may truncate your BigQuery tables to fit the new minimum window requirement.

The 37-month minimum isn’t arbitrary. It aligns with Google’s guidance that advertisers should maintain at least three years of conversion data for Smart Bidding model training, seasonal pattern recognition, and year-over-year performance comparison. For WooCommerce stores, 37 months of event data covers three complete holiday seasons — Black Friday, Christmas, and New Year — which is the minimum dataset Smart Bidding needs to predict seasonal demand shifts accurately.

Checking your current setting takes two minutes. In the Google Cloud console, navigate to BigQuery Data Transfer, select your GA4 transfer configuration, and check the data retention field. If it’s set below 37 months, extend it now. You cannot recover BigQuery data that’s been truncated by a connector update — the deletion is permanent, and GA4’s own interface retains far less historical detail than BigQuery.

Cost implications are real. Extending retention from 14 to 37 months roughly triples the storage volume for GA4 event data. For a WooCommerce store generating 500,000 events per month, that’s the difference between approximately 7 million rows and 18.5 million rows in BigQuery. At BigQuery’s storage pricing of $0.02 per GB per month for active storage, the incremental cost is typically $5–15 per month — far less than the value of the historical data you’d lose.

When Ads Columns Go Null

The Google Ads connector schema change on June 15 replaces column names and data types to match Ads API v23 — existing queries that reference v22 column names will return null values without any error message.

Google Ads API versions deprecate on an 18-month cycle. When a new API version ships, the BigQuery Data Transfer connector for Google Ads eventually adopts the new schema. Ads API v23 changes the schema for date columns, data types, and resource naming conventions — and on June 15, 2026, those changes propagate to your BigQuery Ads transfer tables.

The Google Ads BigQuery connector adopts Ads API v23 schema changes on June 15, 2026, altering date columns and data types in existing transfer tables (Google Cloud, 2026).

Here’s what breaks. A Looker Studio dashboard that references a column called segments_date may find that column renamed to segments.date (dot notation matching the v23 resource hierarchy). A dbt model that casts a date column as STRING may fail when the column type changes to DATE. A custom SQL query that joins on a campaign ID column may produce null joins if the column width or format changes.

Scenario	v22 Schema (Before June 15)	v23 Schema (After June 15)	Impact
Date columns	STRING format date fields	DATE type, dot notation	Queries using string operations on dates break
Resource names	Underscore-separated	Dot-separated hierarchy	Column references return null
New fields	Not present	Added with null defaults	SELECT * queries return extra columns
Deprecated fields	Populated	Removed or renamed	Existing queries lose data

The silent failure mode is the most dangerous part. When a SQL query references a column that no longer exists in the schema, BigQuery doesn’t always throw an error — particularly in views and Looker Studio data sources that use permissive column resolution. The dashboard loads, the chart renders, and the numbers are wrong. The store owner sees their Google Ads cost data looking normal while their conversion value column shows zero — and assumes the campaigns stopped converting.

The Fragile Connector Pattern

Managed connector schemas have changed 14 times in 18 months — every change is a potential reporting break, and the cadence is accelerating as Google ships Ads API versions faster.

The June 2026 changes aren’t isolated incidents. Managed connector schemas have changed 14 times in the past 18 months across GA4 and Google Ads Data Transfer configurations. Each change requires every downstream query, view, dashboard, and model to be audited and potentially updated.

Managed connector schemas have changed 14 times in the past 18 months across GA4 and Google Ads Data Transfer configurations (Google Cloud changelog, 2025–2026).

For WooCommerce stores that built their analytics stack on BigQuery using Google’s built-in connectors, this creates a maintenance burden that scales with complexity. A simple setup — one GA4 property, one Google Ads account, two dashboards — might survive a schema change with 30 minutes of work. A mature setup — multiple properties, multiple ad accounts, custom attribution models, Looker Studio dashboards shared with the team, dbt models feeding into business intelligence — needs hours of audit and testing after each connector change.

The root cause is architectural. When you use Google’s Data Transfer connectors, you accept Google’s schema. You don’t control the column names, the data types, the retention policies, or the update cadence. Google changes any of these, and your reporting breaks. The connector is convenient — it’s a click-to-configure pipeline — but the convenience comes with a dependency that creates ongoing fragility.

This is the same fragility pattern WooCommerce stores face with browser-side tracking. Relying on Google’s managed infrastructure means accepting that the infrastructure changes on Google’s schedule, not yours.

You may be interested in: WooCommerce HPOS 10.7: Why Your Tracking Still Reads Stale Order Data

The Owned Schema Alternative

Streaming events directly to BigQuery through a server-side pipeline gives you a schema you control — one that doesn’t change when Google ships a new API version or updates a connector.

When you own the pipeline that writes to BigQuery, you own the schema. An owned server-side pipeline streams a stable schema you control, with column names that don’t change unless you change them, data types that match your reporting needs, and retention policies set by your business requirements — not by a connector update.

An owned server-side pipeline streams a stable schema you control — eliminating the dependency on Google’s connector schema decisions (Seresa, 2026).

For WooCommerce stores, this means event data — page views, product views, add-to-cart actions, purchases — lands in BigQuery tables with consistent column names across years. When you build a year-over-year comparison dashboard, the columns from January 2025 and January 2027 are identical because your pipeline wrote them both using the same schema.

This doesn’t eliminate every dependency on Google’s connectors. If you need Google Ads campaign performance data — impressions, clicks, cost, quality score — that data originates in Google’s systems and you access it through their APIs or connectors. The Ads connector schema change on June 15 still affects those Ads-specific tables. But your event data, your customer data, and your conversion data — the data your server collects — stays stable.

The Transmute Engine™ pipeline writes event data to BigQuery using a versioned schema that Seresa controls. When Google changes their connector schema, your event tables are unaffected. Your dashboards that blend event data with Ads data need attention only on the Ads side — and even there, the pipeline normalises Ads data into a stable intermediate format before it reaches your reporting layer.

You may be interested in: Ad Blocking Didn’t Peak — 1.77 Billion Users Entrenched in 2026

Key Takeaways

• Two June 2026 deadlines affect BigQuery reporting: GA4 retention window must be 37 months minimum by June 1; Google Ads connector schema changes to Ads API v23 on June 15.
• Ads queries return null, not errors: When column names change, existing queries silently return empty values — dashboards load normally with wrong numbers.
• Managed connectors are structurally fragile: 14 schema changes in 18 months means every downstream query, dashboard, and model needs periodic audit and update.
• Extend GA4 retention now: Check your Data Transfer configuration and set retention to 37+ months before June 1 — truncated BigQuery data cannot be recovered.
• Owned pipelines provide schema stability: Server-side event streaming to BigQuery gives you column names and data types that don’t change on Google’s schedule.

References

• Google Cloud — BigQuery Data Transfer Service release notes (2026). GA4 connector 37-month retention requirement effective June 1, 2026.
• Google Cloud — BigQuery Data Transfer Service changelog (2025–2026). Google Ads connector Ads API v23 schema changes effective June 15, 2026.
• Google Ads API — Version deprecation schedule and schema migration documentation (2026). 18-month API version lifecycle.
• Google Cloud — BigQuery Data Transfer Service change log (2024–2026). 14 managed connector schema changes across GA4 and Google Ads in 18 months.
• Google Cloud — BigQuery storage pricing documentation (2026). Active storage at $0.02 per GB per month.
• Google Ads — Smart Bidding data requirements (2025). Minimum 37 months for seasonal pattern recognition across three holiday cycles.
• Seresa — Server-side event pipeline schema stability analysis (2026). Owned schema vs managed connector dependency comparison.

When Google changes the schema under your dashboards, the stores running owned event pipelines keep their numbers straight. Talk to Seresa about server-side BigQuery pipelines for WooCommerce.