GA4 Attribution Changes in 2026: Why Conversions Look Lower

If your GA4 conversions dropped in April 2026 for no obvious reason, the problem is probably not your tracking. In April, Google quietly reworked GA4 attribution in 2026: the acquisition lookback window shrank from 90 to 30 days, the first-click model was removed, the attribution reports were merged, and every conversion is now attributed independently. No clear notification, no warning banner. As a result, dozens of teams are comparing April 2026 to April 2025 without realizing the counting rules shifted under their feet. This article breaks down the four concrete changes, explains which one causes which anomaly in your reports, and gives you the corrective actions to take now.

The 4 GA4 attribution changes in 2026, at a glance

Before the details, here is the overview. Each row ties a change to the anomaly it creates in your reports, so you can quickly pin down what explains your drop.

Change 1: the lookback window drops from 90 to 30 days

This is the most under-documented change, and likely the main culprit behind your drop. By default, GA4 credited a conversion to acquisition touchpoints that occurred within the previous 90 days. Since April 2026, that default window for acquisition events (such as first_visit and first_open) is cut to 30 days.

In practice, a user who discovers your site through organic search in January and then converts in March is no longer tied to that first visit: the gap exceeds 30 days. The conversion shifts to a more recent touchpoint, or falls to direct if no other channel appears within the window. Long-cycle channels (SEO, awareness, content, consideration campaigns) absorb the loss, while bottom-of-funnel channels (brand, retargeting, direct) appear to perform better than before.

To spot the impact, compare the channel breakdown before and after April at a constant traffic volume. If organic and display shrink while direct swells, with no real change to your campaigns, the lookback window is the most likely cause. You can push the window back to 90 days under Admin, Attribution settings, but the change is not retroactive and affects only future conversions. One caveat: this reduction applies to Analytics reports; the attribution of conversions imported into Google Ads follows its own settings, which you should check separately on the Ads side.

Change 2: the first-click model is gone

Google had announced the end of rules-based models years ago, but first-click remained available for historical analysis. In April 2026, it was permanently removed from the comparison tool. Reports and explorations that relied on it now come up empty or get silently switched to another model.

The question becomes what to use instead. Data-driven attribution (DDA) stays the default and the most defensible model for budget decisions, because it weights each touchpoint by its actual contribution. Last-click keeps its value for accounting reconciliation and control. If your team used first-click to measure discovery and market opening, however, no model replaces it directly: you have to lean on the assisted conversions report or a path analysis to recover that first-touch signal. To allocate budget across channels despite this constraint, our guide on cross-channel budgeting in GA4 lays out a method that holds up when the model changes.

Change 3: independent attribution per conversion

A subtler shift: each key event type is now attributed independently. Previously, several conversions in the same journey could share a common attribution logic; now GA4 computes attribution separately for each conversion.

The effect in your reports is that totals no longer reconcile quite as they used to, which is disorienting at first. Used well, though, this decoupling is an improvement: it lets you analyze the real weight of each channel per objective, for instance separating channels that generate leads from those that drive purchases. Take advantage of it by segmenting your reading by key event rather than reasoning on an aggregate total. Two settings deserve parallel attention, because they distort the count at the source: generate_lead now requires the value and currency parameters to count as a key event (a frequent trap in B2B lead gen), and begin_checkout was dropped from the default key events list, so re-enable it manually if you need it. A GA4 configuration audit lets you verify these points in a few minutes.

Change 4: the unified Attribution report

On the interface side, the former “Conversion paths” and “Model comparison” reports merged into a single Attribution tab. Users accustomed to the old navigation lose their bearings, but the consolidated view makes analysis easier once you get used to it.

Two areas deserve your attention. The assisted conversions view shows channels that intervene upstream without closing the sale: this is where you recover part of the first-touch signal lost with first-click. The Early, Mid and Late analysis positions each channel in the journey, helping you understand whether a channel opens, nurtures or closes. Get into the habit of cross-reading these two views before concluding that a channel “does not convert”: it may be playing an assist role that last-click hides entirely.

The structural limits of data-driven attribution

Since DDA becomes the default fallback, you need to know its blind spots, because they are rarely shown clearly. The model works as a black box: Google does not disclose the weighting applied to each touchpoint, which makes any counter-analysis difficult. It requires a minimum volume, roughly 400 conversions over 28 days for a given conversion type; below that threshold, the property silently falls back to last-click without warning you. Many small properties believe they run on data-driven while they have actually been on last-click for months.

On top of that sit the classic blind spots of browser-side measurement: truncated iOS journeys, poorly stitched cross-device sessions, touchpoints lost when consent is not granted. DDA fixes none of these gaps; it simply attributes what it sees. To rebuild more reliable journeys and check what GA4 hides in the interface, the raw export to BigQuery remains the best option: our BigQuery queries for analyzing GA4 show how to recompute attribution on your own rules.

Checklist: 3 actions to take now

To regain control of your property, three moves are enough today.

1. Check the lookback window. Admin, Attribution settings: decide deliberately between 30 and 90 days based on your cycle length, rather than living with the default.
2. Review your key events. Confirm that generate_lead sends value and currency, and re-enable begin_checkout if your funnel depends on it.
3. Confirm you are actually on DDA. If your property falls below 400 conversions over 28 days, you are effectively on last-click: adjust how you read the reports accordingly.

The key takeaway: GA4 does not actively warn you when its counting rules change. Monitoring these settings is now part of the job. Document the date of each change in your property so that next time you can tell a real performance shift from a mere configuration effect.