On August 11, 2026, Google quietly replaced the fixed 3-day GA4 engaged-view conversion window with an editable 1–30 day range, and turned the six click-through conversion (CTC) presets into any integer from 1 to 90 days. The setting shipped in the What’s new notes with no migration guidance and, critically, no stated position on retroactivity. That means the moment anyone on your team widens a window, your conversion series gets a structural break at the edit date — and every efficiency metric downstream of it (tROAS, tCPA, YoY video reporting) starts lying without telling you.
This is not a “should you care about video measurement” think piece. It’s the exact sequence to change the window safely, rebaseline what breaks, and avoid the failure modes that will otherwise surface in a QBR three months from now.
Why the GA4 engaged-view conversion window change breaks your baselines
An engaged-view conversion credits a conversion when someone watches at least 10 seconds of a skippable YouTube or Google Display Network video ad, does not click, and then converts inside the window. Because there’s no click, EVC is the most lookback-sensitive metric you report: widen the window from 3 to 14 days and you mechanically pull in more conversions with zero change in actual performance.
Three things make this dangerous:
- The discontinuity is silent. Google has not confirmed the change is retroactive, so historical rows likely keep the old 3-day logic while new rows use the edited window. Your time series has a seam that no annotation flags for you.
- It feeds automation. Smart Bidding optimizes to the conversion counts it sees. Change the window and the bidder chases a target that was calibrated on different data.
- It breaks comparability. Any cross-account or YoY comparison that mixes pre- and post-edit windows is now apples-to-oranges.
The 9-step GA4 engaged-view conversion window rebaselining playbook
Step 1 — Freeze config and snapshot baselines before touching anything
Export the current EVC and CTC window settings for every conversion action, plus 13 months of daily conversions by action and campaign. Store it as an immutable snapshot. Effort: 2–3 hours. Failure mode: editing first and reconstructing the old baseline later — you can’t, because the setting doesn’t version itself.
Step 2 — Decide whether to change the window at all
Default to not changing it. The old 3-day window is defensible and stable. Change the window if: your median video-influenced purchase cycle is demonstrably longer than 3 days (considered purchases, B2B, high-AOV DTC) and you have geo holdouts to validate the wider credit. Skip it if: you run short-cycle DTC, you can’t validate incrementally, or your bidding is currently stable and hitting targets. Effort: 1 hour + one stakeholder decision.
Step 3 — Pick a defensible EVC window tied to your purchase cycle
If you change it, anchor the number to data, not vibes. Pull the distribution of time-from-video-exposure to conversion and pick a window near the 75th–80th percentile of genuinely video-influenced conversions. Document the rationale in one paragraph. Effort: half a day. Failure mode: picking 30 days because “more conversions look good” — you’ll inflate EVC credit and mislead budget allocation.
Step 4 — Align CTC windows deliberately, not by default
The same release lets CTC run 1–90 days. Don’t let each conversion action drift to a different window. Standardize CTC windows by conversion type (e.g., 30-day for purchase, 7-day for lead) across accounts so cross-channel ROAS stays comparable. Effort: 2 hours.
Step 5 — Mark the discontinuity everywhere the data lands
Annotate the exact edit date in GA4, in your BI tool, and in the data warehouse (a window_change_flag column keyed to date). Every dashboard that trends EVC must show the seam. Effort: 1–2 hours. Failure mode: trusting a single GA4 annotation nobody sees in the exported tables that feed your models.
Step 6 — Rebaseline dashboards and YoY comparisons
Rebuild video-performance baselines from the edit date forward and clearly separate “3-day-window era” from “new-window era.” For YoY, either restate the prior period to the new window (if you can approximate it) or annotate that YoY is non-comparable until you have 12 months on the new window. Effort: 1–2 days.
Step 7 — Recalibrate bidding targets after counts shift
If conversion counts jump because the window widened, your existing tROAS/tCPA targets are now mis-set. Adjust targets to hold real efficiency constant, then monitor for 2–4 weeks. Move targets gradually — abrupt changes plus a data seam make it impossible to attribute performance swings. Effort: 2 hours setup + 2–4 weeks monitoring. Failure mode: letting Smart Bidding “figure it out” through the discontinuity and misreading the transient as a trend.
Step 8 — Validate EVC credit against incrementality
EVC is a modeled, click-free credit. Before you let a wider window steer budget, run a geo holdout on the video channel and compare incremental lift to the EVC-credited conversions. If EVC materially overstates lift, discount it in planning. As a reference point, Haus found branded search posted a median incremental ROAS of just 0.70x across 225 DTC geo tests — proof that platform-credited conversions and real lift routinely diverge. Effort: 4–6 weeks for a clean geo test.
Step 9 — Lock it down with change-control governance
Put the conversion-window setting under change control: who can edit it, required sign-off, and a logged annotation on every change. Add a monthly audit that diffs live window settings against the documented standard. Effort: half a day + ~30 min/month. Failure mode: a well-meaning account manager “optimizing” the window six months from now and silently reopening the whole problem.
Decision summary
| Situation | Recommended move |
|---|---|
| Short-cycle DTC, stable bidding, no geo-test capability | Keep the 3-day EVC window; lock it under change control |
| Considered purchase / B2B, cycle > 3 days, geo holdouts available | Widen EVC to the 75th–80th percentile of purchase lag; validate with incrementality |
| Multi-account or agency-managed portfolio | Standardize windows by conversion type across accounts before any edits |
FAQ
What is the GA4 engaged-view conversion window?
It’s the lookback period during which a conversion is credited to a video ad after a viewer watches at least 10 seconds of a skippable YouTube or Display video without clicking. As of August 2026 it is editable from 1 to 30 days, replacing the previous fixed 3-day window.
Is the new GA4 conversion window change retroactive?
Google has not confirmed retroactivity. Treat it as non-retroactive: assume historical data keeps the old window and new data uses the edited one, which creates a discontinuity at the edit date. Annotate that date everywhere before you change anything.
How do I change the engaged-view conversion window in GA4?
Go to Advertising → Conversion management → Settings (the more-options menu) and set the EVC window (1–30 days) and CTC window (1–90 days) per conversion action. Snapshot your current settings and baselines first.
Should I widen the conversion window to capture more conversions?
Only if your video-influenced purchase cycle is genuinely longer than 3 days and you can validate the extra credit with a geo holdout. Widening purely to increase reported conversions inflates EVC credit and distorts budget allocation.
Sources
- Google Analytics drops the fixed 3-day engaged-view conversion window — PPC Land
- Google Analytics adds custom conversion attribution windows — Search Engine Land
- About Engaged-view conversions — Google Ads Help
- Google Confirms Sunset Details for 4 Attribution Models — Search Engine Land
- Is Meta’s Incremental Attribution Outperforming Standard Attribution? — Haus
