If your reporting baseline depends on Google Ads history beyond three years, the data has a clock on it now. The Google Ads Data Retention Policy takes effect on June 1, 2026, and the safest response is not panic — it is a quiet, repeatable Google Ads data export habit you can hand off without supervision.
What the new retention policy actually says
The policy page is brief and worth quoting carefully. Beginning June 1, 2026, hourly, daily and weekly reporting data is available for 37 months. Monthly, quarterly and annual data is available for 11 years. Reach and frequency metrics — including unique users and average impression frequency — are available for 3 years as an exception. After those periods, the data will not be accessible via the Google Ads interface or APIs.
Three implications that change how you think about reporting:
- Granular vs. aggregated diverges. Your hourly performance series and your monthly performance series now expire on different schedules. Year-over-year analyses that depend on hourly or daily granularity have the shorter horizon.
- “Not accessible” includes the API. The policy is not a UI quirk — it covers both the Google Ads interface and the APIs. There is no back door once the window closes.
- Reach and frequency is the shortest window. If audience-level reach reporting matters for media plans or board decks, that’s the tightest deadline.
What historical Google Ads exports are meant to protect
Five workflows quietly depend on long historical access today, and they’re the first to fail when the retention window closes on a date you didn’t plan for:
- Annual and multi-year audits, especially when the question is “what changed”.
- Year-over-year campaign and account reporting, including channel-mix comparisons.
- Budget planning that uses seasonality from prior years to project the next.
- Query, category and campaign histories used to defend strategy choices to leadership.
- Change diagnosis — explaining a current anomaly by referring to a comparable historical window.
What is verified about Google Ads API and Scripts
Google describes the Google Ads API as the programmatic interface for managing large or complex Google Ads accounts and campaigns. Google Ads Scripts is the in-account JavaScript automation surface. Two notes that matter for export planning:
- Neither surface is required for every advertiser. Most small accounts can survive on scheduled report downloads from the UI — but those downloads are still subject to the same retention windows above.
- If you choose the API, the query language is GAQL; if you choose Scripts, you stay inside the account’s JavaScript environment. Pick the surface that matches who will own the workflow in three years, not the one that’s most fun to write today.
What to export before you trust a reporting baseline
The dimensions and metrics worth pulling into your own store, at the resolution they expire at:
- Campaign and ad group: spend, impressions, clicks, conversions, conversion value by date segment.
- Search term and query reports where the API or Scripts make them available — these are notoriously hard to reconstruct after the fact.
- Asset, final URL, and landing page dimensions where available — the asset-level history is what defends creative decisions later.
- Conversion action rows separately, so that downstream reattribution doesn’t silently rewrite history.
- Device and geographic segments at the date granularity you depend on.
- Reach and frequency outputs if you run video or display — three years is the shortest fuse.
The export workflow as logic, not pasted code
The actual script will vary by stack, but the logic is the same for any export you intend to run unattended:
- Define a rolling date window. Run for “the last complete calendar period”, not “today” — that avoids partial-day rows and time-zone slippage.
- Query the metrics at the granularity you care about. If you need hourly later, export hourly now; resampling cannot recover detail you didn’t store.
- Write to an append-only target. Google Sheets is fine for small accounts; BigQuery scales better for multi-account histories. The key word is append — never overwrite a previous snapshot.
- Log every failure with a timestamp. An export that silently failed on a Sunday is worse than no export, because it gives you false confidence three quarters later.
- Schedule at a cadence that matches your reporting cycle. Daily is enough for most accounts; some teams add a midday safety run for high-spend accounts. There is no single “correct” cadence — match your data freshness needs.
- Keep the code and the credentials separately reviewed. The export script is small surface area, but it has read access to commercial data.
Reporting-baseline export checklist (table)
This is the table I keep open when scoping a Google Ads retention export with a new client:
| Data area | Why it matters | Export field / source | Risk if missing | Safe action |
|---|---|---|---|---|
| Hourly performance | Day-parting decisions, automated rule debugging | Google Ads API segments.date + segments.hour or Scripts hourly stats | Day-part history disappears after 37 months | Schedule append-only export at hourly granularity now |
| Search term / query | Negative-keyword reasoning, query-mix shifts | search_term_view via API or AdsApp.report in Scripts | Cannot defend negative decisions made years ago | Store with campaign + ad group keys |
| Asset / creative | Creative iteration history | Asset performance reports where available | Lose context for which creative cycle produced lift | Snapshot before campaign or asset structure changes |
| Conversion action | Re-attribution can quietly change historical totals | conversion_action_view + per-action metrics | “Yesterday’s numbers” stop matching “today’s view” | Freeze a per-action snapshot per reporting period |
| Reach & frequency | Media plan validation | video/display reach metrics | 3-year window is the tightest in the policy | Treat as the highest-priority export |
| Monthly aggregates | YoY and 3-year storyboards | API or Scripts monthly roll-ups | 11-year window is generous but not infinite | Annual snapshot at fiscal year close |
Operational checklist before you call it done
- Naming convention: account ID + report name + granularity + run-date in the filename or table name.
- Time zone: store the account time zone alongside the data — UTC-only or local-only both create reconciliation pain later.
- UTM and conversion action consistency: if you renamed a conversion mid-quarter, the export should carry both names.
- Access control: the export destination should be readable only by the people who already have the live account.
- Error logs: a separate “export run log” sheet or table noting success, row count and runtime.
- Backup destination: at least one copy off the Google account that owns the data, so an account suspension does not also lose the history.
- Documentation: a short README explaining how to re-run the export and where the source-of-truth lives.
Assume you are running a three-year year-over-year analysis next quarter — that is the canonical scenario this export is built to survive (this is an explicit hypothetical, not a benchmark of how often any specific account uses the data).
What the export does not fix
An export is a copy, not a correction. Three boundaries to stay honest about:
- Exported data does not fix bad tracking — if a conversion event was mis-fired during the period, the export preserves the wrong number.
- Exported data does not prove causality — week-over-week change is still correlated with everything else you shipped.
- Exported data does not replace consent or tracking validation — pull the export, but keep the live tracking review.
If you want a second pair of eyes on the export plan, the storage choice, or the audit you intend to run against the data, that is what my marketing audit and consulting engagement covers — we walk the retention map, the script logic, and the read-out before anything ages out. Longer notes on Google Ads reporting and AI ad surfaces live in Insights, and the library includes ready-to-fork Google Ads scripts such as Pause Low-ROAS Keywords as a starting point for in-account automation.
Frequently Asked Questions
When exactly does the new Google Ads retention policy start?
The policy applies beginning June 1, 2026, per the Google Ads Help page on data retention. The 37-month, 11-year and 3-year windows are measured from that effective date forward.
What happens to data after the retention window closes?
The Google Ads Help page is unambiguous: after the retention period, the data is no longer accessible via the Google Ads interface or APIs. There is no admin override; if you didn’t export it, you don’t have it.
Should I use Google Ads Scripts or the Google Ads API for the export?
For most single-account workflows, Scripts is fine and stays inside the account’s JavaScript environment. For multi-account, MCC-level or BigQuery-bound workflows, the API plus GAQL is the more durable surface. Pick the surface whoever owns this in three years can actually maintain.
Can I just rely on UI downloads instead of building an export pipeline?
Manual UI downloads are subject to the same retention windows — if the underlying data has aged out, the download will not contain it. UI-only is fine for low-frequency reporting; for anything you’ll defend to leadership in two or three years, an automated append-only export is safer.
Do I need to export everything for every account today?
No. Prioritise the data with the shortest fuse (reach and frequency, three years), the data hardest to reconstruct (search term and asset-level history), and the data you actually use (the dimensions in your real reports). Anything else can wait for a second pass.