Any granular Google Ads data older than 37 months — hourly, daily, weekly breakdowns — is gone as of today.
Not archived. Not paginated behind a paywall. Deleted.
Live: June 1, 2026, across all accounts globally.
What Changed
Google now runs a tiered retention model:
- Granular stats (hour / day / week segmented): 37-month rolling window. Any data before May 2023 is permanently inaccessible as of today.
- Aggregate stats (monthly, quarterly, annual): 11-year retention. Year-over-year comparisons at the summary level survive.
- Reach & frequency metrics: 36-month cap only.
The Google Ads API now returns a DateRangeError for any query requesting granular segments.date data outside the 37-month window. BigQuery Data Transfer backfills for expired dates are disabled — and re-running historical transfers may overwrite existing BigQuery data with null values for the expired range.
Who Gets Hit Hardest
Key findings driving impact:
- Regulated industries (finance, healthcare, legal): FINRA and HIPAA mandates often require 3–7 years of detailed records. That window no longer exists natively in Google Ads.
- Seasonal retailers: Black Friday 2022 daily performance data is gone. Multi-year holiday trend analysis requires a warehouse that was pre-populated before today.
- API and script users: Any custom pipeline querying
segments.datefor ranges > 37 months breaks with a DateRangeError — silently, if your error handling is loose. - GA4 integrations: The Google Analytics Data API truncates joined Google Ads cost, click, and impression metrics to 36 months when combined with granular date dimensions.
The Fix: BigQuery or Bust
Google's path forward is the BigQuery Data Transfer Service (DTS). Data already in BigQuery is not subject to the deletion policy — it persists indefinitely. The catch: there's no retroactive path for data that's already expired.
| Action | Impact | Effort | Owner |
|---|---|---|---|
| Set up BigQuery DTS with append-mode pipeline | Critical — stops all future data loss | Medium | Engineering / Analytics |
| Decouple BI dashboards from direct API calls for dates > 3 years | High — prevents broken reports | Low | Analytics |
| Audit compliance archives for regulated verticals | High — legal exposure | Medium | Finance / Legal |
| Migrate Looker Studio / Power BI to warehouse-backed data sources | Medium | Low | Analytics |
Critical warning: Configure your pipeline in "append" mode — not "refresh." A refresh job that tries to re-pull expired data will overwrite your existing BigQuery records with nulls, compounding the loss.
Root cause: Google's data infrastructure was never designed for indefinite per-account granular retention at scale. This is a supply-side capacity decision dressed as a product policy.
If you don't have a BigQuery pipeline running today, you cannot recover the data that's already been deleted. What you still have — the 37 months still in the window — is exiting every day.
Build the pipeline now.
Founder CEO/CTO
