Meta's 7-Day Click Window vs GA4's 90 Days: Why the Numbers Never Match
Meta says the ad drove the sale. GA4 says organic search did. Both are right, because they are answering different questions. Meta credits on a 7-day click, 1-day view window. GA4 credits on a 90-day lookback with data-driven weights. The numbers were never going to match.
The two windows
Meta's default attribution setting is 7-day click, 1-day view: a purchase counts for the ad if the buyer clicked it within 7 days or viewed it within 1 day. You can change it at the ad set level to 1-day click or 7-day click with 1-day view, but the default is what most accounts run, and it is what most reporting reflects. The view-through part is the important half. Meta credits an ad for a purchase when the buyer saw it and did not click, which GA4 never does.
GA4's default is a 90-day lookback for most key events, with data-driven attribution spreading credit across the touchpoints in the path. A buyer who clicked a Meta ad 10 days ago, then came back through organic search and bought, is a Meta conversion in Meta's window and a shared or organic conversion in GA4's, depending on the model's weights. The same path, the same purchase, two different credit assignments, and neither tool is wrong.
Direct visits add a third wrinkle. GA4 excludes direct visits from credit unless the entire path was direct, which quietly reallocates credit away from the direct bucket. Meta has no equivalent rule, because Meta never sees the direct visit at all. It sees the click and the view, and credits from there.
The iOS hole in the middle
Apple's App Tracking Transparency framework means a large share of iOS users never send Meta the signal that would connect the click to the purchase. Meta fills the gap with modeled conversions, which are statistical estimates, not observed events, and the model is better for some advertisers than others depending on volume and signal quality. GA4 has the same problem from the other side: it only sees the purchases its own tag observed, and the ones it never saw are not in any model. Related reading: Last Click Gives One Channel the Sale and Lies About the Rest.
The result is a reconciliation gap that no model setting closes. One analysis across 16 stores found GA4 recorded a median 76 percent of real orders, and no attribution model setting brings the rest back. The missing quarter is not an attribution problem. It is a measurement problem, and it lives in both tools at once.
Visits with no known source land in Direct in GA4, for example clicks from some apps and AI chats, and those visits then get excluded from credit by the direct rule. A buyer who arrived from an AI chat, clicked nothing trackable, and bought, is a direct visit that GA4 will not credit to the channel that actually drove it, because GA4 never saw that channel.
How to report it without lying
Pick one source of truth and name it. If the CFO gets Meta's number, say it is Meta's number, on a 7-day click, 1-day view window, including modeled conversions. If the CFO gets GA4's number, say it is GA4's number, on a 90-day data-driven lookback. Never present one tool's number with the other tool's definition, and never invoice across tools.
Compare trends within one tool, not levels across tools. Meta's trend tells you whether Meta is getting better or worse at driving purchases. GA4's trend tells you whether the site's overall acquisition is getting better or worse. Blended MER, total revenue divided by total spend, is the check that keeps both honest: if MER is falling while both tools claim improvement, one of them is counting something the other is not. More on this: The Discount Code That Stole Credit From Every Other Channel.
Frequently asked questions
Can I change Meta's attribution window?
Yes, at the ad set level: 1-day click or 7-day click, with 1-day view. The default is 7-day click, 1-day view, and most accounts never change it.
Does GA4 see Meta view-through conversions?
No. GA4 credits clicks, not views. A buyer who saw a Meta ad and bought without clicking is a Meta conversion and invisible to GA4's attribution.
Which number do I report to the CFO?
One source of truth, named explicitly, plus blended MER as the check. Never present Meta's number with GA4's definition, or the other way around.
Why is GA4 missing a quarter of my orders?
Measurement gaps, not attribution settings. GA4 only credits the purchases its tag observed, and iOS opt-outs, ad blockers, and untracked sources all reduce what it sees.