On June 1, 2026, Google quietly changed how long it will let you look backward. Google Ads now retains granular reporting data — the hourly, daily, and weekly segments most analysts live in — for only 37 months. Aggregated monthly, quarterly, and annual data sticks around for 11 years. Nothing about your spend, your conversions, or your bidding changes. What changes is your ability to reconstruct the past at the resolution you are used to.
For most consumer advertisers this is a shrug. For B2B SaaS accounts — long sales cycles, heavy seasonality, budgets justified to a board with year-over-year trend charts — it quietly removes a tool you may not have noticed you depended on. This post covers exactly what changed, why 37 months is the number that bites SaaS specifically, which reports to export first, and how to do it before the oldest data crosses the line.
What changed on June 1, 2026
Google announced the policy on the Google Ads Developer Blog: granular performance statistics are now retained for 37 months, while higher-level aggregates are retained for 11 years. The enforcement mechanism is blunt. Per the PPC News Feed summary of the announcement, "queries requesting granular segments (e.g., daily or weekly data) older than 37 months will return a DateRangeError.INVALID_DATE error." That error surfaces through the API, and the same windows constrain what the UI will render in date-segmented reports.
The 37-month boundary is a rolling window, not a one-time purge. Every day, another day of granular history drops off the back. So this is not an event you handle once on June 1 and forget — it is a standing condition. If your account is three years old today, you already sit right at the edge: the daily detail from your first months of spend is now either gone or about to go. Google's own guidance, in its developer note, is terse — advertisers are advised to "update their queries to adapt to these changes and export necessary data before the deadline." That second clause is the whole job.
Why 37 months is the number that bites SaaS
Three years sounds generous until you map it against how B2B SaaS actually plans. SaaS demand is seasonal in ways that only show up at weekly resolution — the pre-renewal Q4 surge, the January budget-reset spike, the August enterprise lull. Building a credible forecast for next Q1 means laying this Q1 against the last two or three, ideally at the week level so you can see the shape of the ramp, not just the monthly total. A 37-month granular window gives you roughly three prior cycles before the detail vanishes — and the oldest of those is already eroding.
The longer your sales cycle, the worse the squeeze. If a click in March becomes a closed deal in September, the loop you care about spans two and three quarters before you can even score it. Tying granular click-and-cost history to deals that closed much later — the backbone of honest CAC and payback analysis — depends on having both ends of that loop available at the same resolution. When the cost side ages out of granular reporting at 37 months, the long-cycle cohorts are exactly the ones you lose first. The accounts that most need a long memory are the ones the policy trims hardest.
The reports you'll lose first — and which to export
Not everything needs archiving. Monthly and annual aggregates are safe for 11 years, so high-level spend and conversion totals are not at risk. The exposure is entirely in the day- and week-segmented views. Prioritize the granular reports that feed decisions you actually revisit: daily campaign and ad-group performance (cost, clicks, conversions, conversion value), daily search-term reports, and daily geographic and device splits if you bid or budget on them. These are the cuts that let you answer "what did the ramp look like last time" — and they are precisely what the 37-month window removes.
Pay special attention to anything that underpins attribution and board-level metrics. If you report blended CAC, payback, or channel contribution off granular history, archive the daily data behind those numbers now — otherwise next year's YoY chart will have a hole where the oldest comparison period used to be. The same logic applies to the metrics your leadership tracks: see our breakdown of the five Google Ads metrics SaaS boards care about and how different attribution models read the same spend differently. If a number ends up in a board deck, its granular source data should live somewhere you control.
How to export before it ages out
The durable answer is a scheduled export rather than a panicked download. Set up a recurring pull of daily-segmented reports into a warehouse — BigQuery is the natural fit given the native Google Ads Data Transfer integration, but any warehouse you query works. Use the Google Ads API or Google Ads Scripts to extract daily campaign, ad-group, keyword, and search-term reports and land them where Google's retention policy cannot reach. Once that pipeline runs nightly, the 37-month limit stops mattering: your own copy keeps accumulating indefinitely at full granularity.
For accounts without engineering time to spare, a one-time backstop buys you breathing room. Export the last several years of day-segmented reports to CSV from the UI, or run a single Google Ads Script that dumps daily data across all campaigns to a Google Sheet or cloud storage before the oldest rows cross the line. It is less elegant than a warehouse, but a static archive of granular history is far better than discovering the gap next budget season. Whichever route you take, do it now — every day you wait, another day of the oldest detail is permanently out of reach.
What stays, and how to keep trend analysis honest
It helps to be precise about what you have not lost. The 11-year retention on monthly, quarterly, and annual aggregates means your long-run spend and conversion totals are intact — you can still pull a clean monthly time series back the better part of a decade. Billing and invoice history is unaffected. What you give up is resolution on the oldest periods: you can see that April 2022 cost a given amount, you just can't break that April into its 30 days inside Google Ads anymore.
Practically, that means rebuilding any analysis that silently assumed infinite granular history. Year-over-year week-level comparisons should now draw their older arm from your exported archive, not from a live Google Ads query that will eventually error out. Audit your reporting stack for queries that request daily or weekly segments across multi-year ranges and point them at your warehouse copy. Folding this check into a recurring routine — alongside the rest of your conversion tracking hygiene — keeps the policy from surprising you a year from now when a board chart quietly comes up short.
What this is not — and the one-hour action plan
Do not confuse this with the consent and tracking changes that also landed in 2026. This is a reporting-retention policy, full stop. Your conversion tags still fire, enhanced conversions still match, and your bid strategies still optimize on exactly the same signal they did in May. Nothing about live measurement degrades. The only capability that narrows is how far back you can interrogate granular history inside Google Ads — which is an archival problem, not a measurement one, and archival problems are cheap to solve if you solve them early.
The one-hour version: stand up (or schedule) a daily-segment export of campaign, keyword, and search-term reports to a warehouse or CSV; confirm it captures at least the last 37 months before the oldest data ages out; and repoint any multi-year week-level reports at that archive. If you want a second set of eyes on whether your account's reporting and attribution will survive the cutover cleanly, our Google Ads audit includes a review of exactly which historical reports your decisions depend on and whether they're safely archived. The data is still there today. In 37 months, the granular version won't be.