Google removed the fixed three-day lookback window on engaged-view conversions in Google Analytics on August 11, 2026, replacing it with an editable range of 1 to 30 days and simultaneously converting the six preset options for click-through conversions into any integer between 1 and 90 days.

The change was documented in the "What's new in Google Analytics" release notes, the same Help Center surface where Google has published most of its measurement changes this year. There was no blog post and no press release. According to the documentation, "Google Analytics conversions now support custom integer lookback windows for click-through conversions (CTC) and engaged-view conversions (EVC)."

Two settings changed. Neither is cosmetic.

What the release note specifies

For engaged-view conversions, the release note states that advertisers can "Edit EVC conversion windows to any integer from 1 to 30 days (previously fixed at 3 days)." The parenthetical carries the weight. Until August 11, the three-day figure was not a default that could be adjusted; it was a constant applied to every property regardless of the sales cycle behind it.

For click-through conversions, the note states that windows can be set "to any integer from 1 to 90 days (previously limited to preset values of 1, 7, 14, 30, 60, or 90 days)." The outer boundary is unchanged. What disappears is the ladder. An advertiser whose repeat purchase interval sat at 21 days previously chose between 14 and 30, and the choice systematically distorted the reported figure in one direction or the other.

Configuration sits in two places. According to the documentation, custom conversion windows are set "under Advertising > Conversion management > more options icon > Settings, as well as in the linked Google Ads conversion management interface." That dual location matters for accounts where the Analytics property and the Google Ads account are managed by different teams, since the same value is now reachable from either side of the link.

The stated rationale is brief. The update, according to Google, "allows Google Analytics to align conversion attribution lookback windows more closely to your unique business cycles."

Three days was never a neutral number

Engaged-view conversions count a viewer who watched a qualifying portion of a video advertisement without clicking, then converted afterwards. The counting rules are specific: skippable in-stream formats require at least 10 seconds of viewing, or the full duration when the advertisement runs shorter, while in-feed and Shorts placements require 5 seconds. Measuring that behaviour has always depended on the Google Ads and Analytics link being in place, because the conversion itself is recorded on the advertiser's property while the view is recorded on YouTube.

The three-day boundary was the constraint that defined how much of that behaviour ever surfaced. A viewer who watched a 30-second brand film, considered the purchase for a week, and then bought directly produced no engaged-view conversion at all. The purchase was recorded, but the video that preceded it received no credit. For categories with short consideration cycles, three days was defensible. For furniture, insurance, enterprise software or anything financed, it functioned as a systematic undercount of video contribution.

That asymmetry had a documented history inside Google's own products. Analysis of GA4 attribution settings noted that engaged views of YouTube advertisements always carried three-day lookback windows regardless of the property-level configuration, producing properties in which different interaction types operated on different attribution clocks simultaneously. Advertisers who lengthened their key event lookback window found the video window unmoved. As of August 11, that particular divergence is configurable.

The commercial reading is not complicated. Video budgets have been justified against a metric whose ceiling was set three days after the view. Raising the ceiling to 30 days will raise the count. Whether the additional conversions represent incremental effect or simply reassign credit that last-click models were already awarding to search and direct traffic is a question the release note does not touch.

The preset ladder and what replaces it

The six-step click-through ladder was a rounding mechanism. Every account was pushed toward one of six numbers, and the distance between adjacent steps grew as the values rose: 30 to 60 is a doubling, and 60 to 90 adds half again. An advertiser with a 45-day cycle had no accurate option.

Integer granularity removes that. It also removes an excuse. A property reporting on a 30-day window when the underlying purchase cycle runs to 38 days can no longer attribute the gap to platform limitations.

Retroactivity is the unresolved variable. The release note is silent on whether adjusting a window re-computes historical figures or applies only forward. Prior documentation of GA4 attribution behaviour has held that lookback window changes do not operate retroactively and apply from the moment of the change. If that behaviour persists, any account that widens a window will show a discontinuity in its reported conversion series at the date of the edit, with the period before the change counted under the old rule and the period after counted under the new one. Year-on-year comparisons drawn across that boundary would compare two different definitions.

Nothing in the August 11 note confirms or contradicts that. The absence is itself information for anyone maintaining dashboards built on conversion counts.

A year of conversion plumbing

The August 11 update lands inside a sequence rather than on its own. On January 16, 2026, Google made conversion attribution settings adjustable independently for every conversion, so that a lead form submission and an e-commerce transaction inside the same property could carry different models and different windows. Attribution model options under that change included data-driven, last click, first click, linear, time decay and position-based. The August update extends the same logic from the model to the clock: per-conversion settings are of limited use if the window under them can only take six values.

Cross-channel conversion reporting reached the Google Analytics Data API in alpha on May 4, 2026, adding a ConversionSpec field that filters by conversion action IDs and by attribution model, with DATA_DRIVEN and LAST_CLICK as the accepted values. Developers pulling that data now face a variable that can differ per conversion action and, from August 11, per integer day.

The rest of the year has moved in the same direction. Enhanced conversions for web and for leads collapsed into a single toggle, removing method selection from the interface. A new Source Group dimension arrived on June 11 to consolidate fragmented source values across Facebook, Instagram and TikTok. Campaign data import began requiring a currency field on every cost upload from July 28. A diagnostic launched on July 30 to flag properties where GBRAID and gad_ parameters are missing from advertisement click URLs, a gap that silently reclassifies paid sessions as organic. A campaign data import validation report followed on August 10, one day before the conversion window change.

Read as a group, these are repairs to a measurement layer that had accumulated fixed assumptions. Each removes a constraint that was defensible when set and had stopped being defensible.

Why this matters for advertisers and publishers

Conversion windows are not a reporting preference. They are an input to automated bidding, and the number of conversions a campaign reports determines how much the system is willing to pay for the next one.

Google's own migration guidance for advertisers moving off standalone Display campaigns recommended setting the conversion attribution window to more than 28 days when rebuilding inside Demand Gen. Under the preset ladder, "more than 28 days" resolved to 30, 60 or 90. It now resolves to any of 62 values. For an account running lift studies or migrating campaign structures, the difference between 30 and 35 days is no longer a rounding decision imposed by the interface.

The bidding context is immediate. Google is applying a change to how bidding targets are enforced from August 17, 2026, a change that some advertisers calculated would require doubling stated target CPAs to hold current delivery. Conversion volume feeds directly into that calculation. Altering a lookback window six days before a bidding enforcement change alters the denominator of every efficiency metric an advertiser might use to judge the result, and separating the two effects afterwards will be difficult for accounts that change both.

There is a cross-platform dimension as well. Meta moved in the opposite direction, eliminating the 7-day and 28-day view-through attribution windows from its Ads Insights API with effect from January 12, 2026, retaining click and one-day view options only. Meta then redefined click-through attribution to count only link clicks for website and in-store conversions in March 2026. One platform is narrowing the definitions available; the other is widening the range of configurable values. Marketing teams comparing YouTube against paid social now do so across measurement systems whose flexibility is diverging, not converging.

For publishers and agencies reporting to clients, the practical consequence is that two properties measuring identical behaviour can now produce conversion counts that differ by any integer margin of window length. Comparability across accounts, already weak, weakens further. The release note offers no default recommendation and no guidance on selecting a value, which places the burden of justifying the number entirely on whoever sets it.

The measurement confidence problem this sits inside is not new. Research cited in coverage of the Meridian integration into Analytics 360 found 54.1% of marketers reporting no year-on-year improvement in measurement confidence despite continued tooling investment. Granular conversion windows add precision to a configuration surface. Whether precision at the configuration layer translates into confidence at the reporting layer is a separate question, and one that a release note published without a blog post does not attempt to answer.

Timeline

Summary

Who: Google, through the Google Analytics product team, affecting advertisers, agencies and analysts operating Google Analytics properties, particularly those with linked Google Ads accounts running YouTube video campaigns where engaged-view conversions are counted.

What: Conversion lookback windows became configurable as custom integers. Engaged-view conversion windows, previously fixed at three days for every property, can now be set to any integer from 1 to 30 days. Click-through conversion windows, previously limited to the preset values 1, 7, 14, 30, 60 and 90 days, can now be set to any integer from 1 to 90 days. The outer 90-day ceiling for click-through conversions is unchanged; what changed is the granularity beneath it.

When: August 11, 2026, published in the "What's new in Google Analytics" release notes. The change arrived one day after the campaign data import validation report and six days before the August 17 bidding target enforcement change in Google Ads.

Where: Inside Google Analytics under Advertising > Conversion management > more options icon > Settings, and equivalently within the linked Google Ads conversion management interface.

Why: According to Google, the update allows conversion attribution lookback windows to align more closely with individual business cycles. The practical effect is that the three-day constraint on video-driven conversions, which systematically excluded consideration periods longer than 72 hours, no longer applies, and the six-step ladder that forced click-through windows into rounded values has been removed. Because conversion counts feed automated bidding, changes to these settings alter both reported performance and the signals that bidding systems optimise against.