Google Ads Cuts Granular Data Window From 11 Years to 37 Months

Google is cutting granular Google Ads reporting access from 11 years down to 37 months, a 72 percent reduction in the addressable historical window for daily, hourly, and weekly performance data. The reversal lands less than 18 months after the 11-year policy was put in place, and roughly seven months after it actually took effect on November 13, 2024. For any team that built attribution models, year-over-year benchmarks, or BigQuery pipelines on the assumption of long-range Google Ads history, June 1, 2026 is now a hard deadline.

The Numbers

The headline contraction is 11 years to 37 months for granular data, but the actual surface area of the change is wider than a single number suggests. As PPC Land reported, the May 1, 2026 notice on the Google Ads Developer Blog, authored by Nadine Wang of the Advertising and Measurement APIs Team, applies the new window to the Google Ads API, Google Ads scripts, the Google Analytics Data API, and the BigQuery Data Transfer Service. A parallel Help Center update with the same May 1, 2026 effective date confirms the limit applies to the Google Ads interface itself, not just developer surfaces.

The 37-month ceiling is specifically scoped to granular performance statistics broken down at daily, hourly, and weekly levels. Higher aggregations at monthly, quarterly, and yearly granularity continue to retain 11 years. So the policy is better described as a tiered retention model than a flat cut: granular = 37 months, aggregated = 11 years. That distinction matters because it constrains the type of analysis that's possible at long range, not the existence of historical data.

One sub-category is even tighter. Reach and frequency metrics, including unique users, average impression frequency per user, the 7-day and 30-day frequency variants, and frequency distribution thresholds at 1+, 2+, 3+, 4+, 5+, and 10+, will be accessible for only three years. That's roughly one month shorter than the general 37-month window, and shorter still than the 36-month truncation point applied in the Google Analytics Data API.

For comparison, the source notes that Facebook Ads retains data for approximately 37 months, so Google is effectively converging on a parity point Meta already occupies. The DV360 API and CM360 API, by contrast, are unchanged at a 24-month retention period. So inside Google's own ad stack you now have three different retention regimes: 24 months for DV360/CM360, 37 months granular plus 11 years aggregated for Google Ads, and a separate three-year window for reach and frequency. The source does not disclose why DV360 and CM360 were exempted, which matters because it suggests the constraint is specific to Google Ads infrastructure rather than a uniform policy choice across Google's ad business.

What's Actually New

The technical behavior is where the announcement gets sharp, and where engineering teams need to focus. The Google Ads API and Google Ads scripts will return a hard DateRangeError.INVALID_DATE error starting June 1, 2026 for any query that requests granular segments such as segments.date or segments.week for date ranges older than 37 months. Future API versions are slated to return a more specific DateRangeError.REQUESTED_DATE_GRANULARITY_NOT_SUPPORTED error. To pull anything older, queries must use segments.month, segments.quarter, or segments.year, and unsegmented historical queries have to align precisely with calendar month boundaries to succeed.

The Google Analytics Data API takes a different approach: silent truncation rather than an error. When the dimension date (or date-equivalents like day or Day of week) is present in a report, affected metrics are truncated to the latest 36-month window. The affected metrics are Advertiser Ad Cost, Clicks, and Impressions. Reports only get truncated if they include all three affected metrics, cover a date range of 37 months or older, and include the date dimension.

This divergence is the genuinely new operational risk. One API throws; the other quietly clips. A reporting pipeline that joins Google Ads API data with Google Analytics Data API data could end up with mismatched row counts and no thrown exception to flag the discrepancy. Anyone running unified dashboards on top of both surfaces should be auditing for this before June 2026.

The BigQuery Data Transfer Service has its own asymmetry. For Google Ads and Search Ads 360 connectors, backfill runs with dates older than 37 months stop populating new data, but data already transferred to BigQuery tables stays put. For GA4, however, a manual transfer triggered for a report date 37 months or older will overwrite existing data in the BigQuery table with an empty value. That's a destructive operation gated behind an operator action that previously did the opposite of destroying data. The unanswered question here, the source does not specify, is whether scheduled (non-manual) GA4 transfers carry the same overwrite risk or whether the destructive behavior is bounded only to manual triggers. The blast radius depends entirely on that distinction.

What's Priced In for Performance Marketing

Performance marketing teams have been watching Google compress retention windows for over a year, so the direction of travel is not surprising. In February 2025, Google set a 540-day limit on Customer Match data retention across Google Ads and Display & Video 360, effective April 7, 2025. In June 2025, Google AdMob cut Ads Activity report data to seven years and slashed User Activity report data to just 90 days. The 37-month Google Ads ceiling fits that pattern.

What's not priced in, in my read, is the speed and reversal. Google announced the 11-year retention policy in October 2024 and put it into effect on November 13, 2024. The 37-month ceiling was published less than seven months after that policy took live effect. Teams that took the 11-year commitment at face value and built long-horizon attribution models, marketing mix modeling inputs, or year-over-year benchmarks against the documented retention window now have to redesign those systems on a roughly 13-month timeline. That's an unusually short half-life for a stated retention policy from a vendor of Google's scale.

The other under-priced piece is the cost shift to advertisers. If granular history older than 37 months is needed for trend modeling, MMM training, or seasonality decomposition, the storage burden moves to the customer. Teams that haven't been exporting Google Ads data to their own warehouses now have a 13-month window to build an export pipeline before the long tail becomes irretrievable.

Contrarian View

The consensus reading is that Google is taking something away. The contrarian read: most advertisers don't actually use granular Google Ads data older than 37 months, and the 11-year policy was always more of a marketing posture than an operational reality. Day-level campaign data from 2022 has limited analytical value in 2026 because the auction dynamics, audience definitions, and even the campaign types have changed enough that point-in-time comparisons are noisy at best.

If that's right, the meaningful constraint isn't the 37-month cap, it's the reach and frequency three-year window and the specific GA4 BigQuery overwrite risk. Those affect live workflows. The granular cap mostly affects reports nobody runs.

I'd qualify that with a caveat: the source does not provide usage statistics on how often granular queries beyond 37 months are actually issued, so the bound here is qualitative. If Google had numbers showing meaningful long-range query volume, they'd presumably have used them to defend the 11-year policy rather than reverse it.

Key Takeaways

• Granular Google Ads reporting drops from 11 years to 37 months on June 1, 2026, with reach and frequency metrics capped at three years and aggregated monthly/quarterly/yearly data still retained for 11 years.
• The Google Ads API throws DateRangeError.INVALID_DATE, the Google Analytics Data API silently truncates to 36 months, and the GA4 BigQuery Data Transfer Service can overwrite existing historical data with empty values on manual backfill triggers. Three different failure modes on three surfaces.
• The 37-month window puts Google Ads roughly at parity with Facebook Ads retention, while DV360 and CM360 stay at 24 months under the unchanged Google Ads API documentation umbrella.
• Teams should audit any Google Ads API query, Google Ads script, or BigQuery Data Transfer pipeline that requests day-level or week-level data older than 37 months, and migrate to segments.month, segments.quarter, or segments.year with calendar-aligned ranges before June 1, 2026.
• Testable prediction: if this policy holds, we should see a measurable spike in Google Ads to BigQuery custom export volumes in Q4 2025 and Q1 2026 as advertisers race to snapshot historical granular data before the deadline. If that spike doesn't materialize, it confirms the contrarian read that nobody was actually using the long tail.