Meta Says 4.2x. Your Bank Account Disagrees.
Three 2026 changes stack up against Meta-reported ROAS: Meta's July change to off-platform data handling, Safari's wider tracking protection in iOS 26, and Shopify's new default for app pixels.
Three platforms, three answers, one budget to defend. Your GA4 export already holds the answer, and the first finding is free and never leaves your browser. Get your first finding free
By Joris van Huët, Founder & CEOUpdated
Each one changes what the dashboards can see, not what your ads caused. The fix is not another pixel: it is an independent causal read of what actually moved revenue, from your GA4 export.
What changed in 2026
Meta removed the off-platform data opt-out in July (the EU is excluded for now), so people can no longer disconnect the activity businesses share with Meta from their accounts. Safari's tracking protection, widened in iOS 26, removes click identifiers such as fbclid from links opened in private browsing, Mail and Messages, so the click that used to stitch an ad to a purchase does not always arrive. Shopify's January 2026 default for app pixels, “Optimized”, can pause what a third-party pixel receives. Each change moves what the platforms can see; together they detach reported ROAS from what actually happened.
The size of the gap, in your numbers
Published estimates of how far platform-reported ROAS runs above true contribution use different methods, and we have not found one that is independently audited, so we do not quote one. Your own gap is measurable: take a month, add up the revenue every ad platform claims, and set it against what your store actually took. If the platforms together claim more than you sold, the difference is credit counted twice. If your dashboard looks better than your bank account, this is why.
MER vs ROAS: the sanity check
Marketing efficiency ratio, total revenue over total ad spend, cannot be gamed by attribution. If platform ROAS says 4.2x while MER says 1.8x and falling, the platform number is not describing your business. MER tells you something is wrong; it cannot tell you which channel to fix. That requires channel-level causal measurement.
The independent second opinion
Platforms grade their own homework, and every incentive points toward claiming more. A causal read runs on your GA4 export, models what revenue would have happened anyway, and returns what each channel caused, next to last-click. Run it next to your platform dashboards and budget on the difference.
Measure it, then decide
The method behind the read: causal attribution, incrementality testing, and marketing mix modeling, priced at €99 one-time or €299/mo.
FAQ
Why did my Meta ROAS drop in 2026?
Reported ROAS moved because measurement changed, not necessarily your ads: Meta's off-platform data change, Safari's wider tracking protection in iOS 26, and Shopify's new default for app pixels all landed within months of each other.Is Meta-reported ROAS inflated?
Often, because the platform can only see part of the picture and credits itself generously inside it. How much varies by store, and we have not found an independently audited figure, so we do not quote one. Measure yours: set the revenue Meta reports against a causal read of what Meta actually caused.What is MER vs ROAS?
MER is total revenue divided by total ad spend, immune to attribution tricks but blind to channels. ROAS is per-channel but relies on attribution. Use MER as the sanity check and causal measurement for the channel answer.How do I verify what Meta actually caused?
Run an independent causal read on your GA4 export. Causality Engine shows what each channel caused, next to what last-click gave it, in 1 to 2 minutes, €99 one-time, refundable within 30 days, no questions asked.