Google Ads Is Cutting Historical Data Access to 37 Months. Export Before June 1.

What Google is removing

On June 1, 2026, Google Ads will cut the data retention window for granular reporting from what was effectively unlimited — over a decade of queryable history — to 37 months. Granular here means day-level, hour-level, and week-level breakdowns across campaigns, ad groups, and keywords. Reach and frequency data gets cut separately to 36 months.

That's not a housekeeping update. That's 2014 through early 2023 disappearing from your account on a fixed calendar date.

Google announced this in March with minimal fanfare. No dashboard notification, no direct email to account owners. You probably heard about it from a PPC newsletter or haven't heard about it at all.

What this affects in practice

The cut applies to the reports most used for seasonal modeling and multi-year benchmarking: campaign performance, ad group performance, keyword data, auction insights — all of it queried at daily or weekly time grains. Anything older than 37 months will return empty results after June 1.

High-level aggregated data, like monthly totals in the account overview, may persist longer under different retention rules. But if your workflow involves pulling day-level cost, clicks, or conversion data across quarters to build a Q4-over-Q4 comparison going back to 2020, that specific data is what's getting deleted.

For seasonality modeling, this is a real constraint. Most solid seasonal models run on 3–5 years of same-period comparisons. A 37-month window leaves you three complete calendar years — barely enough for a confident one-year forward projection.

Who gets hurt most

Long-running accounts with evergreen campaigns have the most to lose. If your Google Ads account predates 2023 and you've never exported the raw daily data, you're holding historical signal that Google is about to erase.

Agencies managing multiple accounts should treat this as a client-by-client data task, not an internal note to revisit. The window is 24 days and the exports aren't fast.

Attribution and measurement teams are also affected. Reach and frequency metrics feed marketing mix models and upper-funnel incrementality analysis. Losing the granular historical view disrupts any model that depends on extended time series — and most good ones do. If you use Google's Meridian tool or any external MMM, the older granular data is part of what gives those models their explanatory power.

What to export before June 1

Google Ads Editor doesn't retain historical data — it reads from the live account. The export has to happen through the Ads web interface or the API.

From the Reports tab, you can schedule bulk downloads of campaign and keyword performance. At minimum: daily breakdowns going back to 2020, keyword-level data for your highest-spend campaigns, auction insights for your core branded and non-branded terms, and device performance splits if you've been tracking mobile/desktop differences over time.

If your account has conversion actions that changed definitions, value rules, or attribution windows during that period, note the dates when changes happened. Exported data without that context is hard to interpret later.

BigQuery export is the cleaner long-term solution. Enabling it now won't recover data already past the retention window, but it creates a continuous export going forward. Set it up alongside a one-time historical pull before June 1, and you're covered on both ends.

One thing worth checking: some third-party reporting tools that connect to the Google Ads API may have already been pulling and storing your data. If you've been running Looker Studio dashboards or any BI connector for years, those datasets may already contain the history you need. Verify before assuming you need to start from scratch.

What this actually tells you about platform dependency

Google is deleting data it costs money to store. That's the full explanation. The 90-day notice period is roughly the minimum required to avoid advertiser backlash.

Advertisers who have treated Google Ads as a system of record — pulling data when needed rather than warehousing it continuously — are learning the same lesson that attribution teams learned during iOS 14: data you don't control can disappear. The platforms are not your data partners. They're your data landlords.

The accounts in the best position right now are the ones that set up BigQuery exports or third-party data warehousing years ago and don't have to scramble. That setup isn't complicated, but it requires someone having decided to do it before a deadline appeared.

If you want to see where your account stands on measurement, data setup, and tracking configuration before June 1, Gromerce's free audit surfaces those gaps in about three minutes.

Your Google Ads history goes back further than 37 months. June 1 is the last day to do anything about it.

Sources: PPC.land, Search Engine Land, May 2026